xref: /netbsd-src/sys/kern/init_sysctl.c (revision a6f3f22f245acb8ee3bbf6871d7dce989204fa97)
1 /*	$NetBSD: init_sysctl.c,v 1.209 2015/08/25 14:52:31 pooka 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 1.209 2015/08/25 14:52:31 pooka Exp $");
34 
35 #include "opt_sysv.h"
36 #include "opt_compat_netbsd.h"
37 #include "opt_modular.h"
38 #include "pty.h"
39 
40 #include <sys/types.h>
41 #include <sys/param.h>
42 #include <sys/sysctl.h>
43 #include <sys/cpu.h>
44 #include <sys/errno.h>
45 #include <sys/systm.h>
46 #include <sys/kernel.h>
47 #include <sys/unistd.h>
48 #include <sys/disklabel.h>
49 #include <sys/cprng.h>
50 #include <sys/vnode.h>
51 #include <sys/mount.h>
52 #include <sys/namei.h>
53 #include <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/resource.h>
60 #include <sys/resourcevar.h>
61 #include <sys/exec.h>
62 #include <sys/conf.h>
63 #include <sys/device.h>
64 #include <sys/stat.h>
65 #include <sys/kauth.h>
66 #include <sys/ktrace.h>
67 #include <sys/ksem.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_rtc_offset(SYSCTLFN_PROTO);
115 static int sysctl_kern_maxproc(SYSCTLFN_PROTO);
116 static int sysctl_kern_hostid(SYSCTLFN_PROTO);
117 static int sysctl_kern_defcorename(SYSCTLFN_PROTO);
118 static int sysctl_kern_cptime(SYSCTLFN_PROTO);
119 #if NPTY > 0
120 static int sysctl_kern_maxptys(SYSCTLFN_PROTO);
121 #endif /* NPTY > 0 */
122 static int sysctl_kern_lwp(SYSCTLFN_PROTO);
123 static int sysctl_kern_forkfsleep(SYSCTLFN_PROTO);
124 static int sysctl_kern_root_partition(SYSCTLFN_PROTO);
125 static int sysctl_kern_drivers(SYSCTLFN_PROTO);
126 static int sysctl_security_setidcore(SYSCTLFN_PROTO);
127 static int sysctl_security_setidcorename(SYSCTLFN_PROTO);
128 static int sysctl_kern_cpid(SYSCTLFN_PROTO);
129 static int sysctl_hw_usermem(SYSCTLFN_PROTO);
130 static int sysctl_hw_cnmagic(SYSCTLFN_PROTO);
131 
132 static void fill_lwp(struct lwp *l, struct kinfo_lwp *kl);
133 
134 /*
135  * ********************************************************************
136  * section 1: setup routines
137  * ********************************************************************
138  * These functions are stuffed into a link set for sysctl setup
139  * functions. They're never called or referenced from anywhere else.
140  * ********************************************************************
141  */
142 
143 /*
144  * this setup routine is a replacement for kern_sysctl()
145  */
146 SYSCTL_SETUP(sysctl_kern_setup, "sysctl kern subtree setup")
147 {
148 	extern int kern_logsigexit;	/* defined in kern/kern_sig.c */
149 	extern fixpt_t ccpu;		/* defined in kern/kern_synch.c */
150 	extern int dumponpanic;		/* defined in kern/subr_prf.c */
151 	const struct sysctlnode *rnode;
152 
153 	sysctl_createv(clog, 0, NULL, NULL,
154 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
155 		       CTLTYPE_INT, "maxvnodes",
156 		       SYSCTL_DESCR("Maximum number of vnodes"),
157 		       sysctl_kern_maxvnodes, 0, NULL, 0,
158 		       CTL_KERN, KERN_MAXVNODES, CTL_EOL);
159 	sysctl_createv(clog, 0, NULL, NULL,
160 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
161 		       CTLTYPE_INT, "maxproc",
162 		       SYSCTL_DESCR("Maximum number of simultaneous processes"),
163 		       sysctl_kern_maxproc, 0, NULL, 0,
164 		       CTL_KERN, KERN_MAXPROC, CTL_EOL);
165 	sysctl_createv(clog, 0, NULL, NULL,
166 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
167 		       CTLTYPE_INT, "maxfiles",
168 		       SYSCTL_DESCR("Maximum number of open files"),
169 		       NULL, 0, &maxfiles, 0,
170 		       CTL_KERN, KERN_MAXFILES, CTL_EOL);
171 	sysctl_createv(clog, 0, NULL, NULL,
172 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
173 		       CTLTYPE_INT, "argmax",
174 		       SYSCTL_DESCR("Maximum number of bytes of arguments to "
175 				    "execve(2)"),
176 		       NULL, ARG_MAX, NULL, 0,
177 		       CTL_KERN, KERN_ARGMAX, CTL_EOL);
178 	sysctl_createv(clog, 0, NULL, NULL,
179 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE|CTLFLAG_HEX,
180 		       CTLTYPE_INT, "hostid",
181 		       SYSCTL_DESCR("System host ID number"),
182 		       sysctl_kern_hostid, 0, NULL, 0,
183 		       CTL_KERN, KERN_HOSTID, CTL_EOL);
184 	sysctl_createv(clog, 0, NULL, NULL,
185 		       CTLFLAG_PERMANENT,
186 		       CTLTYPE_STRUCT, "vnode",
187 		       SYSCTL_DESCR("System vnode table"),
188 		       sysctl_kern_vnode, 0, NULL, 0,
189 		       CTL_KERN, KERN_VNODE, CTL_EOL);
190 #ifndef GPROF
191 	sysctl_createv(clog, 0, NULL, NULL,
192 		       CTLFLAG_PERMANENT,
193 		       CTLTYPE_NODE, "profiling",
194 		       SYSCTL_DESCR("Profiling information (not available)"),
195 		       sysctl_notavail, 0, NULL, 0,
196 		       CTL_KERN, KERN_PROF, CTL_EOL);
197 #endif
198 	sysctl_createv(clog, 0, NULL, NULL,
199 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
200 		       CTLTYPE_INT, "posix1version",
201 		       SYSCTL_DESCR("Version of ISO/IEC 9945 (POSIX 1003.1) "
202 				    "with which the operating system attempts "
203 				    "to comply"),
204 		       NULL, _POSIX_VERSION, NULL, 0,
205 		       CTL_KERN, KERN_POSIX1, CTL_EOL);
206 	sysctl_createv(clog, 0, NULL, NULL,
207 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
208 		       CTLTYPE_INT, "ngroups",
209 		       SYSCTL_DESCR("Maximum number of supplemental groups"),
210 		       NULL, NGROUPS_MAX, NULL, 0,
211 		       CTL_KERN, KERN_NGROUPS, CTL_EOL);
212 	sysctl_createv(clog, 0, NULL, NULL,
213 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
214 		       CTLTYPE_INT, "job_control",
215 		       SYSCTL_DESCR("Whether job control is available"),
216 		       NULL, 1, NULL, 0,
217 		       CTL_KERN, KERN_JOB_CONTROL, CTL_EOL);
218 	sysctl_createv(clog, 0, NULL, NULL,
219 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
220 		       CTLTYPE_INT, "saved_ids",
221 		       SYSCTL_DESCR("Whether POSIX saved set-group/user ID is "
222 				    "available"), NULL,
223 #ifdef _POSIX_SAVED_IDS
224 		       1,
225 #else /* _POSIX_SAVED_IDS */
226 		       0,
227 #endif /* _POSIX_SAVED_IDS */
228 		       NULL, 0, CTL_KERN, KERN_SAVED_IDS, CTL_EOL);
229 	sysctl_createv(clog, 0, NULL, NULL,
230 		       CTLFLAG_PERMANENT|CTLFLAG_HEX,
231 		       CTLTYPE_INT, "boothowto",
232 		       SYSCTL_DESCR("Flags from boot loader"),
233 		       NULL, 0, &boothowto, sizeof(boothowto),
234 		       CTL_KERN, CTL_CREATE, CTL_EOL);
235 	sysctl_createv(clog, 0, NULL, NULL,
236 		       CTLFLAG_PERMANENT,
237 		       CTLTYPE_STRUCT, "boottime",
238 		       SYSCTL_DESCR("System boot time"),
239 		       NULL, 0, &boottime, sizeof(boottime),
240 		       CTL_KERN, KERN_BOOTTIME, CTL_EOL);
241 	sysctl_createv(clog, 0, NULL, NULL,
242 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
243 		       CTLTYPE_INT, "maxpartitions",
244 		       SYSCTL_DESCR("Maximum number of partitions allowed per "
245 				    "disk"),
246 		       NULL, MAXPARTITIONS, NULL, 0,
247 		       CTL_KERN, KERN_MAXPARTITIONS, CTL_EOL);
248 	sysctl_createv(clog, 0, NULL, NULL,
249 		       CTLFLAG_PERMANENT,
250 		       CTLTYPE_STRUCT, "timex", NULL,
251 		       sysctl_notavail, 0, NULL, 0,
252 		       CTL_KERN, KERN_TIMEX, CTL_EOL);
253 	sysctl_createv(clog, 0, NULL, NULL,
254 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
255 		       CTLTYPE_INT, "rtc_offset",
256 		       SYSCTL_DESCR("Offset of real time clock from UTC in "
257 				    "minutes"),
258 		       sysctl_kern_rtc_offset, 0, &rtc_offset, 0,
259 		       CTL_KERN, KERN_RTC_OFFSET, CTL_EOL);
260 	sysctl_createv(clog, 0, NULL, NULL,
261 		       CTLFLAG_PERMANENT,
262 		       CTLTYPE_STRING, "root_device",
263 		       SYSCTL_DESCR("Name of the root device"),
264 		       sysctl_root_device, 0, NULL, 0,
265 		       CTL_KERN, KERN_ROOT_DEVICE, CTL_EOL);
266 	sysctl_createv(clog, 0, NULL, NULL,
267 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
268 		       CTLTYPE_INT, "fsync",
269 		       SYSCTL_DESCR("Whether the POSIX 1003.1b File "
270 				    "Synchronization Option is available on "
271 				    "this system"),
272 		       NULL, 1, NULL, 0,
273 		       CTL_KERN, KERN_FSYNC, CTL_EOL);
274 	sysctl_createv(clog, 0, NULL, NULL,
275 		       CTLFLAG_PERMANENT,
276 		       CTLTYPE_NODE, "ipc",
277 		       SYSCTL_DESCR("SysV IPC options"),
278 		       NULL, 0, NULL, 0,
279 		       CTL_KERN, KERN_SYSVIPC, CTL_EOL);
280 	sysctl_createv(clog, 0, NULL, NULL,
281 		       CTLFLAG_PERMANENT|CTLFLAG_READONLY,
282 		       CTLTYPE_INT, "sysvmsg",
283 		       SYSCTL_DESCR("System V style message support available"),
284 		       NULL, 0, &kern_has_sysvmsg, sizeof(int),
285 		       CTL_KERN, KERN_SYSVIPC, KERN_SYSVIPC_MSG, CTL_EOL);
286 	sysctl_createv(clog, 0, NULL, NULL,
287 		       CTLFLAG_PERMANENT|CTLFLAG_READONLY,
288 		       CTLTYPE_INT, "sysvsem",
289 		       SYSCTL_DESCR("System V style semaphore support "
290 				    "available"),
291 		       NULL, 0, &kern_has_sysvsem, sizeof(int),
292 		       CTL_KERN, KERN_SYSVIPC, KERN_SYSVIPC_SEM, CTL_EOL);
293 	sysctl_createv(clog, 0, NULL, NULL,
294 		       CTLFLAG_PERMANENT|CTLFLAG_READONLY,
295 		       CTLTYPE_INT, "sysvshm",
296 		       SYSCTL_DESCR("System V style shared memory support "
297 				    "available"),
298 		       NULL, 0, &kern_has_sysvshm, sizeof(int),
299 		       CTL_KERN, KERN_SYSVIPC, KERN_SYSVIPC_SHM, CTL_EOL);
300 	sysctl_createv(clog, 0, NULL, NULL,
301 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
302 		       CTLTYPE_INT, "synchronized_io",
303 		       SYSCTL_DESCR("Whether the POSIX 1003.1b Synchronized "
304 				    "I/O Option is available on this system"),
305 		       NULL, 1, NULL, 0,
306 		       CTL_KERN, KERN_SYNCHRONIZED_IO, CTL_EOL);
307 	sysctl_createv(clog, 0, NULL, NULL,
308 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
309 		       CTLTYPE_INT, "iov_max",
310 		       SYSCTL_DESCR("Maximum number of iovec structures per "
311 				    "process"),
312 		       NULL, IOV_MAX, NULL, 0,
313 		       CTL_KERN, KERN_IOV_MAX, CTL_EOL);
314 	sysctl_createv(clog, 0, NULL, NULL,
315 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
316 		       CTLTYPE_INT, "mapped_files",
317 		       SYSCTL_DESCR("Whether the POSIX 1003.1b Memory Mapped "
318 				    "Files Option is available on this system"),
319 		       NULL, 1, NULL, 0,
320 		       CTL_KERN, KERN_MAPPED_FILES, CTL_EOL);
321 	sysctl_createv(clog, 0, NULL, NULL,
322 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
323 		       CTLTYPE_INT, "memlock",
324 		       SYSCTL_DESCR("Whether the POSIX 1003.1b Process Memory "
325 				    "Locking Option is available on this "
326 				    "system"),
327 		       NULL, 1, NULL, 0,
328 		       CTL_KERN, KERN_MEMLOCK, CTL_EOL);
329 	sysctl_createv(clog, 0, NULL, NULL,
330 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
331 		       CTLTYPE_INT, "memlock_range",
332 		       SYSCTL_DESCR("Whether the POSIX 1003.1b Range Memory "
333 				    "Locking Option is available on this "
334 				    "system"),
335 		       NULL, 1, NULL, 0,
336 		       CTL_KERN, KERN_MEMLOCK_RANGE, CTL_EOL);
337 	sysctl_createv(clog, 0, NULL, NULL,
338 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
339 		       CTLTYPE_INT, "memory_protection",
340 		       SYSCTL_DESCR("Whether the POSIX 1003.1b Memory "
341 				    "Protection Option is available on this "
342 				    "system"),
343 		       NULL, 1, NULL, 0,
344 		       CTL_KERN, KERN_MEMORY_PROTECTION, CTL_EOL);
345 	sysctl_createv(clog, 0, NULL, NULL,
346 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
347 		       CTLTYPE_INT, "login_name_max",
348 		       SYSCTL_DESCR("Maximum login name length"),
349 		       NULL, LOGIN_NAME_MAX, NULL, 0,
350 		       CTL_KERN, KERN_LOGIN_NAME_MAX, CTL_EOL);
351 	sysctl_createv(clog, 0, NULL, NULL,
352 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
353 		       CTLTYPE_STRING, "defcorename",
354 		       SYSCTL_DESCR("Default core file name"),
355 		       sysctl_kern_defcorename, 0, defcorename, MAXPATHLEN,
356 		       CTL_KERN, KERN_DEFCORENAME, CTL_EOL);
357 	sysctl_createv(clog, 0, NULL, NULL,
358 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
359 		       CTLTYPE_INT, "logsigexit",
360 		       SYSCTL_DESCR("Log process exit when caused by signals"),
361 		       NULL, 0, &kern_logsigexit, 0,
362 		       CTL_KERN, KERN_LOGSIGEXIT, CTL_EOL);
363 	sysctl_createv(clog, 0, NULL, NULL,
364 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
365 		       CTLTYPE_INT, "fscale",
366 		       SYSCTL_DESCR("Kernel fixed-point scale factor"),
367 		       NULL, FSCALE, NULL, 0,
368 		       CTL_KERN, KERN_FSCALE, CTL_EOL);
369 	sysctl_createv(clog, 0, NULL, NULL,
370 		       CTLFLAG_PERMANENT,
371 		       CTLTYPE_INT, "ccpu",
372 		       SYSCTL_DESCR("Scheduler exponential decay value"),
373 		       NULL, 0, &ccpu, 0,
374 		       CTL_KERN, KERN_CCPU, CTL_EOL);
375 	sysctl_createv(clog, 0, NULL, NULL,
376 		       CTLFLAG_PERMANENT,
377 		       CTLTYPE_STRUCT, "cp_time",
378 		       SYSCTL_DESCR("Clock ticks spent in different CPU states"),
379 		       sysctl_kern_cptime, 0, NULL, 0,
380 		       CTL_KERN, KERN_CP_TIME, CTL_EOL);
381 	sysctl_createv(clog, 0, NULL, NULL,
382 		       CTLFLAG_PERMANENT,
383 		       CTLTYPE_STRUCT, "consdev",
384 		       SYSCTL_DESCR("Console device"),
385 		       sysctl_consdev, 0, NULL, sizeof(dev_t),
386 		       CTL_KERN, KERN_CONSDEV, CTL_EOL);
387 #if NPTY > 0
388 	sysctl_createv(clog, 0, NULL, NULL,
389 		       CTLFLAG_PERMANENT,
390 		       CTLTYPE_INT, "maxptys",
391 		       SYSCTL_DESCR("Maximum number of pseudo-ttys"),
392 		       sysctl_kern_maxptys, 0, NULL, 0,
393 		       CTL_KERN, KERN_MAXPTYS, CTL_EOL);
394 #endif /* NPTY > 0 */
395 	sysctl_createv(clog, 0, NULL, NULL,
396 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
397 		       CTLTYPE_INT, "maxphys",
398 		       SYSCTL_DESCR("Maximum raw I/O transfer size"),
399 		       NULL, MAXPHYS, NULL, 0,
400 		       CTL_KERN, KERN_MAXPHYS, CTL_EOL);
401 	sysctl_createv(clog, 0, NULL, NULL,
402 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
403 		       CTLTYPE_INT, "monotonic_clock",
404 		       SYSCTL_DESCR("Implementation version of the POSIX "
405 				    "1003.1b Monotonic Clock Option"),
406 		       /* XXX _POSIX_VERSION */
407 		       NULL, _POSIX_MONOTONIC_CLOCK, NULL, 0,
408 		       CTL_KERN, KERN_MONOTONIC_CLOCK, CTL_EOL);
409 	sysctl_createv(clog, 0, NULL, NULL,
410 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
411 		       CTLTYPE_INT, "labelsector",
412 		       SYSCTL_DESCR("Sector number containing the disklabel"),
413 		       NULL, LABELSECTOR, NULL, 0,
414 		       CTL_KERN, KERN_LABELSECTOR, CTL_EOL);
415 	sysctl_createv(clog, 0, NULL, NULL,
416 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
417 		       CTLTYPE_INT, "labeloffset",
418 		       SYSCTL_DESCR("Offset of the disklabel within the "
419 				    "sector"),
420 		       NULL, LABELOFFSET, NULL, 0,
421 		       CTL_KERN, KERN_LABELOFFSET, CTL_EOL);
422 	sysctl_createv(clog, 0, NULL, NULL,
423 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
424 		       CTLTYPE_INT, "labelusesmbr",
425 		       SYSCTL_DESCR("disklabel is inside MBR partition"),
426 		       NULL, LABELUSESMBR, NULL, 0,
427 		       CTL_KERN, CTL_CREATE, CTL_EOL);
428 	sysctl_createv(clog, 0, NULL, NULL,
429 		       CTLFLAG_PERMANENT,
430 		       CTLTYPE_NODE, "lwp",
431 		       SYSCTL_DESCR("System-wide LWP information"),
432 		       sysctl_kern_lwp, 0, NULL, 0,
433 		       CTL_KERN, KERN_LWP, CTL_EOL);
434 	sysctl_createv(clog, 0, NULL, NULL,
435 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
436 		       CTLTYPE_INT, "forkfsleep",
437 		       SYSCTL_DESCR("Milliseconds to sleep on fork failure due "
438 				    "to process limits"),
439 		       sysctl_kern_forkfsleep, 0, NULL, 0,
440 		       CTL_KERN, KERN_FORKFSLEEP, CTL_EOL);
441 	sysctl_createv(clog, 0, NULL, NULL,
442 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
443 		       CTLTYPE_INT, "posix_threads",
444 		       SYSCTL_DESCR("Version of IEEE Std 1003.1 and its "
445 				    "Threads option to which the system "
446 				    "attempts to conform"),
447 		       /* XXX _POSIX_VERSION */
448 		       NULL, _POSIX_THREADS, NULL, 0,
449 		       CTL_KERN, KERN_POSIX_THREADS, CTL_EOL);
450 	sysctl_createv(clog, 0, NULL, NULL,
451 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
452 		       CTLTYPE_INT, "posix_semaphores",
453 		       SYSCTL_DESCR("Version of IEEE Std 1003.1 and its "
454 				    "Semaphores option to which the system "
455 				    "attempts to conform"), NULL,
456 		       200112, NULL, 0,
457 		       CTL_KERN, KERN_POSIX_SEMAPHORES, CTL_EOL);
458 	sysctl_createv(clog, 0, NULL, NULL,
459 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
460 		       CTLTYPE_INT, "posix_barriers",
461 		       SYSCTL_DESCR("Version of IEEE Std 1003.1 and its "
462 				    "Barriers option to which the system "
463 				    "attempts to conform"),
464 		       /* XXX _POSIX_VERSION */
465 		       NULL, _POSIX_BARRIERS, NULL, 0,
466 		       CTL_KERN, KERN_POSIX_BARRIERS, CTL_EOL);
467 	sysctl_createv(clog, 0, NULL, NULL,
468 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
469 		       CTLTYPE_INT, "posix_timers",
470 		       SYSCTL_DESCR("Version of IEEE Std 1003.1 and its "
471 				    "Timers option to which the system "
472 				    "attempts to conform"),
473 		       /* XXX _POSIX_VERSION */
474 		       NULL, _POSIX_TIMERS, NULL, 0,
475 		       CTL_KERN, KERN_POSIX_TIMERS, CTL_EOL);
476 	sysctl_createv(clog, 0, NULL, NULL,
477 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
478 		       CTLTYPE_INT, "posix_spin_locks",
479 		       SYSCTL_DESCR("Version of IEEE Std 1003.1 and its Spin "
480 				    "Locks option to which the system attempts "
481 				    "to conform"),
482 		       /* XXX _POSIX_VERSION */
483 		       NULL, _POSIX_SPIN_LOCKS, NULL, 0,
484 		       CTL_KERN, KERN_POSIX_SPIN_LOCKS, CTL_EOL);
485 	sysctl_createv(clog, 0, NULL, NULL,
486 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
487 		       CTLTYPE_INT, "posix_reader_writer_locks",
488 		       SYSCTL_DESCR("Version of IEEE Std 1003.1 and its "
489 				    "Read-Write Locks option to which the "
490 				    "system attempts to conform"),
491 		       /* XXX _POSIX_VERSION */
492 		       NULL, _POSIX_READER_WRITER_LOCKS, NULL, 0,
493 		       CTL_KERN, KERN_POSIX_READER_WRITER_LOCKS, CTL_EOL);
494 	sysctl_createv(clog, 0, NULL, NULL,
495 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
496 		       CTLTYPE_INT, "dump_on_panic",
497 		       SYSCTL_DESCR("Perform a crash dump on system panic"),
498 		       NULL, 0, &dumponpanic, 0,
499 		       CTL_KERN, KERN_DUMP_ON_PANIC, CTL_EOL);
500 #ifdef DIAGNOSTIC
501 	sysctl_createv(clog, 0, NULL, NULL,
502 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
503 		       CTLTYPE_INT, "panic_now",
504 		       SYSCTL_DESCR("Trigger a panic"),
505 		       sysctl_kern_trigger_panic, 0, NULL, 0,
506 		       CTL_KERN, CTL_CREATE, CTL_EOL);
507 #endif
508 	sysctl_createv(clog, 0, NULL, NULL,
509 		       CTLFLAG_PERMANENT,
510 		       CTLTYPE_INT, "root_partition",
511 		       SYSCTL_DESCR("Root partition on the root device"),
512 		       sysctl_kern_root_partition, 0, NULL, 0,
513 		       CTL_KERN, KERN_ROOT_PARTITION, CTL_EOL);
514 	sysctl_createv(clog, 0, NULL, NULL,
515 		       CTLFLAG_PERMANENT,
516 		       CTLTYPE_STRUCT, "drivers",
517 		       SYSCTL_DESCR("List of all drivers with block and "
518 				    "character device numbers"),
519 		       sysctl_kern_drivers, 0, NULL, 0,
520 		       CTL_KERN, KERN_DRIVERS, CTL_EOL);
521 	sysctl_createv(clog, 0, NULL, NULL,
522 		       CTLFLAG_PERMANENT,
523 		       CTLTYPE_STRUCT, "cp_id",
524 		       SYSCTL_DESCR("Mapping of CPU number to CPU id"),
525 		       sysctl_kern_cpid, 0, NULL, 0,
526 		       CTL_KERN, KERN_CP_ID, CTL_EOL);
527 	sysctl_createv(clog, 0, NULL, &rnode,
528 		       CTLFLAG_PERMANENT,
529 		       CTLTYPE_NODE, "coredump",
530 		       SYSCTL_DESCR("Coredump settings."),
531 		       NULL, 0, NULL, 0,
532 		       CTL_KERN, CTL_CREATE, CTL_EOL);
533 	sysctl_createv(clog, 0, &rnode, &rnode,
534 		       CTLFLAG_PERMANENT,
535 		       CTLTYPE_NODE, "setid",
536 		       SYSCTL_DESCR("Set-id processes' coredump settings."),
537 		       NULL, 0, NULL, 0,
538 		       CTL_CREATE, CTL_EOL);
539 	sysctl_createv(clog, 0, &rnode, NULL,
540 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
541 		       CTLTYPE_INT, "dump",
542 		       SYSCTL_DESCR("Allow set-id processes to dump core."),
543 		       sysctl_security_setidcore, 0, &security_setidcore_dump,
544 		       sizeof(security_setidcore_dump),
545 		       CTL_CREATE, CTL_EOL);
546 	sysctl_createv(clog, 0, &rnode, NULL,
547 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
548 		       CTLTYPE_STRING, "path",
549 		       SYSCTL_DESCR("Path pattern for set-id coredumps."),
550 		       sysctl_security_setidcorename, 0,
551 		       security_setidcore_path,
552 		       sizeof(security_setidcore_path),
553 		       CTL_CREATE, CTL_EOL);
554 	sysctl_createv(clog, 0, &rnode, NULL,
555 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
556 		       CTLTYPE_INT, "owner",
557 		       SYSCTL_DESCR("Owner id for set-id processes' cores."),
558 		       sysctl_security_setidcore, 0, &security_setidcore_owner,
559 		       0,
560 		       CTL_CREATE, CTL_EOL);
561 	sysctl_createv(clog, 0, &rnode, NULL,
562 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
563 		       CTLTYPE_INT, "group",
564 		       SYSCTL_DESCR("Group id for set-id processes' cores."),
565 		       sysctl_security_setidcore, 0, &security_setidcore_group,
566 		       0,
567 		       CTL_CREATE, CTL_EOL);
568 	sysctl_createv(clog, 0, &rnode, NULL,
569 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
570 		       CTLTYPE_INT, "mode",
571 		       SYSCTL_DESCR("Mode for set-id processes' cores."),
572 		       sysctl_security_setidcore, 0, &security_setidcore_mode,
573 		       0,
574 		       CTL_CREATE, CTL_EOL);
575 	sysctl_createv(clog, 0, NULL, NULL,
576 		       CTLFLAG_IMMEDIATE|CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
577 		       CTLTYPE_INT, "no_sa_support",
578 		       SYSCTL_DESCR("0 if the kernel supports SA, otherwise "
579 		       "it doesn't"),
580 		       NULL, 1, NULL, 0,
581 		       CTL_KERN, CTL_CREATE, CTL_EOL);
582 	sysctl_createv(clog, 0, NULL, NULL,
583 			CTLFLAG_PERMANENT,
584 			CTLTYPE_STRING, "configname",
585 			SYSCTL_DESCR("Name of config file"),
586 			NULL, 0, __UNCONST(kernel_ident), 0,
587 			CTL_KERN, CTL_CREATE, CTL_EOL);
588 	sysctl_createv(clog, 0, NULL, NULL,
589 			CTLFLAG_PERMANENT,
590 			CTLTYPE_STRING, "buildinfo",
591 			SYSCTL_DESCR("Information from build environment"),
592 			NULL, 0, __UNCONST(buildinfo), 0,
593 			CTL_KERN, CTL_CREATE, CTL_EOL);
594 
595 	/* kern.posix. */
596 	sysctl_createv(clog, 0, NULL, &rnode,
597 			CTLFLAG_PERMANENT,
598 			CTLTYPE_NODE, "posix",
599 			SYSCTL_DESCR("POSIX options"),
600 			NULL, 0, NULL, 0,
601 			CTL_KERN, CTL_CREATE, CTL_EOL);
602 	sysctl_createv(clog, 0, &rnode, NULL,
603 			CTLFLAG_PERMANENT | CTLFLAG_READWRITE,
604 			CTLTYPE_INT, "semmax",
605 			SYSCTL_DESCR("Maximal number of semaphores"),
606 			NULL, 0, &ksem_max, 0,
607 			CTL_CREATE, CTL_EOL);
608 }
609 
610 SYSCTL_SETUP(sysctl_hw_misc_setup, "sysctl hw subtree misc setup")
611 {
612 
613 	sysctl_createv(clog, 0, NULL, NULL,
614 		       CTLFLAG_PERMANENT,
615 		       CTLTYPE_INT, "usermem",
616 		       SYSCTL_DESCR("Bytes of non-kernel memory"),
617 		       sysctl_hw_usermem, 0, NULL, 0,
618 		       CTL_HW, HW_USERMEM, CTL_EOL);
619 	sysctl_createv(clog, 0, NULL, NULL,
620 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE|CTLFLAG_HEX,
621 		       CTLTYPE_STRING, "cnmagic",
622 		       SYSCTL_DESCR("Console magic key sequence"),
623 		       sysctl_hw_cnmagic, 0, NULL, CNS_LEN,
624 		       CTL_HW, HW_CNMAGIC, CTL_EOL);
625 	sysctl_createv(clog, 0, NULL, NULL,
626 		       CTLFLAG_PERMANENT,
627 		       CTLTYPE_QUAD, "usermem64",
628 		       SYSCTL_DESCR("Bytes of non-kernel memory"),
629 		       sysctl_hw_usermem, 0, NULL, 0,
630 		       CTL_HW, HW_USERMEM64, CTL_EOL);
631 }
632 
633 #ifdef DEBUG
634 /*
635  * Debugging related system variables.
636  */
637 struct ctldebug /* debug0, */ /* debug1, */ debug2, debug3, debug4;
638 struct ctldebug debug5, debug6, debug7, debug8, debug9;
639 struct ctldebug debug10, debug11, debug12, debug13, debug14;
640 struct ctldebug debug15, debug16, debug17, debug18, debug19;
641 static struct ctldebug *debugvars[CTL_DEBUG_MAXID] = {
642 	&debug0, &debug1, &debug2, &debug3, &debug4,
643 	&debug5, &debug6, &debug7, &debug8, &debug9,
644 	&debug10, &debug11, &debug12, &debug13, &debug14,
645 	&debug15, &debug16, &debug17, &debug18, &debug19,
646 };
647 
648 /*
649  * this setup routine is a replacement for debug_sysctl()
650  *
651  * note that it creates several nodes per defined debug variable
652  */
653 SYSCTL_SETUP(sysctl_debug_setup, "sysctl debug subtree setup")
654 {
655 	struct ctldebug *cdp;
656 	char nodename[20];
657 	int i;
658 
659 	/*
660 	 * two ways here:
661 	 *
662 	 * the "old" way (debug.name -> value) which was emulated by
663 	 * the sysctl(8) binary
664 	 *
665 	 * the new way, which the sysctl(8) binary was actually using
666 
667 	 node	debug
668 	 node	debug.0
669 	 string debug.0.name
670 	 int	debug.0.value
671 	 int	debug.name
672 
673 	 */
674 
675 	for (i = 0; i < CTL_DEBUG_MAXID; i++) {
676 		cdp = debugvars[i];
677 		if (cdp->debugname == NULL || cdp->debugvar == NULL)
678 			continue;
679 
680 		snprintf(nodename, sizeof(nodename), "debug%d", i);
681 		sysctl_createv(clog, 0, NULL, NULL,
682 			       CTLFLAG_PERMANENT|CTLFLAG_HIDDEN,
683 			       CTLTYPE_NODE, nodename, NULL,
684 			       NULL, 0, NULL, 0,
685 			       CTL_DEBUG, i, CTL_EOL);
686 		sysctl_createv(clog, 0, NULL, NULL,
687 			       CTLFLAG_PERMANENT|CTLFLAG_HIDDEN,
688 			       CTLTYPE_STRING, "name", NULL,
689 			       /*XXXUNCONST*/
690 			       NULL, 0, __UNCONST(cdp->debugname), 0,
691 			       CTL_DEBUG, i, CTL_DEBUG_NAME, CTL_EOL);
692 		sysctl_createv(clog, 0, NULL, NULL,
693 			       CTLFLAG_PERMANENT|CTLFLAG_HIDDEN,
694 			       CTLTYPE_INT, "value", NULL,
695 			       NULL, 0, cdp->debugvar, 0,
696 			       CTL_DEBUG, i, CTL_DEBUG_VALUE, CTL_EOL);
697 		sysctl_createv(clog, 0, NULL, NULL,
698 			       CTLFLAG_PERMANENT,
699 			       CTLTYPE_INT, cdp->debugname, NULL,
700 			       NULL, 0, cdp->debugvar, 0,
701 			       CTL_DEBUG, CTL_CREATE, CTL_EOL);
702 	}
703 }
704 #endif /* DEBUG */
705 
706 /*
707  * ********************************************************************
708  * section 2: private node-specific helper routines.
709  * ********************************************************************
710  */
711 
712 #ifdef DIAGNOSTIC
713 static int
714 sysctl_kern_trigger_panic(SYSCTLFN_ARGS)
715 {
716 	int newtrig, error;
717 	struct sysctlnode node;
718 
719 	newtrig = 0;
720 	node = *rnode;
721 	node.sysctl_data = &newtrig;
722 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
723 	if (error || newp == NULL)
724 		return (error);
725 
726 	if (newtrig != 0)
727 		panic("Panic triggered");
728 
729 	return (error);
730 }
731 #endif
732 
733 /*
734  * sysctl helper routine for kern.maxvnodes.  Drain vnodes if
735  * new value is lower than desiredvnodes and then calls reinit
736  * routines that needs to adjust to the new value.
737  */
738 static int
739 sysctl_kern_maxvnodes(SYSCTLFN_ARGS)
740 {
741 	int error, new_vnodes, old_vnodes, new_max;
742 	struct sysctlnode node;
743 
744 	new_vnodes = desiredvnodes;
745 	node = *rnode;
746 	node.sysctl_data = &new_vnodes;
747 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
748 	if (error || newp == NULL)
749 		return (error);
750 
751 	/*
752 	 * sysctl passes down unsigned values, require them
753 	 * to be positive
754 	 */
755 	if (new_vnodes <= 0)
756 		return (EINVAL);
757 
758 	/* Limits: 75% of kmem and physical memory. */
759 	new_max = calc_cache_size(vmem_size(kmem_arena, VMEM_FREE|VMEM_ALLOC),
760 	    75, 75) / VNODE_COST;
761 	if (new_vnodes > new_max)
762 		new_vnodes = new_max;
763 
764 	old_vnodes = desiredvnodes;
765 	desiredvnodes = new_vnodes;
766 	error = vfs_drainvnodes(new_vnodes);
767 	if (error) {
768 		desiredvnodes = old_vnodes;
769 		return (error);
770 	}
771 	vfs_reinit();
772 	nchreinit();
773 
774 	return (0);
775 }
776 
777 /*
778  * sysctl helper routine for rtc_offset - set time after changes
779  */
780 static int
781 sysctl_kern_rtc_offset(SYSCTLFN_ARGS)
782 {
783 	struct timespec ts, delta;
784 	int error, new_rtc_offset;
785 	struct sysctlnode node;
786 
787 	new_rtc_offset = rtc_offset;
788 	node = *rnode;
789 	node.sysctl_data = &new_rtc_offset;
790 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
791 	if (error || newp == NULL)
792 		return (error);
793 
794 	if (kauth_authorize_system(l->l_cred, KAUTH_SYSTEM_TIME,
795 	    KAUTH_REQ_SYSTEM_TIME_RTCOFFSET,
796 	    KAUTH_ARG(new_rtc_offset), NULL, NULL))
797 		return (EPERM);
798 	if (rtc_offset == new_rtc_offset)
799 		return (0);
800 
801 	/* if we change the offset, adjust the time */
802 	nanotime(&ts);
803 	delta.tv_sec = 60 * (new_rtc_offset - rtc_offset);
804 	delta.tv_nsec = 0;
805 	timespecadd(&ts, &delta, &ts);
806 	rtc_offset = new_rtc_offset;
807 	return (settime(l->l_proc, &ts));
808 }
809 
810 /*
811  * sysctl helper routine for kern.maxproc. Ensures that the new
812  * values are not too low or too high.
813  */
814 static int
815 sysctl_kern_maxproc(SYSCTLFN_ARGS)
816 {
817 	int error, nmaxproc;
818 	struct sysctlnode node;
819 
820 	nmaxproc = maxproc;
821 	node = *rnode;
822 	node.sysctl_data = &nmaxproc;
823 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
824 	if (error || newp == NULL)
825 		return (error);
826 
827 	if (nmaxproc < 0 || nmaxproc >= PID_MAX)
828 		return (EINVAL);
829 #ifdef __HAVE_CPU_MAXPROC
830 	if (nmaxproc > cpu_maxproc())
831 		return (EINVAL);
832 #endif
833 	maxproc = nmaxproc;
834 
835 	return (0);
836 }
837 
838 /*
839  * sysctl helper function for kern.hostid. The hostid is a long, but
840  * we export it as an int, so we need to give it a little help.
841  */
842 static int
843 sysctl_kern_hostid(SYSCTLFN_ARGS)
844 {
845 	int error, inthostid;
846 	struct sysctlnode node;
847 
848 	inthostid = hostid;  /* XXX assumes sizeof int <= sizeof long */
849 	node = *rnode;
850 	node.sysctl_data = &inthostid;
851 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
852 	if (error || newp == NULL)
853 		return (error);
854 
855 	hostid = (unsigned)inthostid;
856 
857 	return (0);
858 }
859 
860 /*
861  * sysctl helper routine for kern.defcorename. In the case of a new
862  * string being assigned, check that it's not a zero-length string.
863  * (XXX the check in -current doesn't work, but do we really care?)
864  */
865 static int
866 sysctl_kern_defcorename(SYSCTLFN_ARGS)
867 {
868 	int error;
869 	char *newcorename;
870 	struct sysctlnode node;
871 
872 	newcorename = PNBUF_GET();
873 	node = *rnode;
874 	node.sysctl_data = &newcorename[0];
875 	memcpy(node.sysctl_data, rnode->sysctl_data, MAXPATHLEN);
876 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
877 	if (error || newp == NULL) {
878 		goto done;
879 	}
880 
881 	/*
882 	 * when sysctl_lookup() deals with a string, it's guaranteed
883 	 * to come back nul terminated. So there.  :)
884 	 */
885 	if (strlen(newcorename) == 0) {
886 		error = EINVAL;
887 	} else {
888 		memcpy(rnode->sysctl_data, node.sysctl_data, MAXPATHLEN);
889 		error = 0;
890 	}
891 done:
892 	PNBUF_PUT(newcorename);
893 	return error;
894 }
895 
896 /*
897  * sysctl helper routine for kern.cp_time node. Adds up cpu time
898  * across all cpus.
899  */
900 static int
901 sysctl_kern_cptime(SYSCTLFN_ARGS)
902 {
903 	struct sysctlnode node = *rnode;
904 	uint64_t *cp_time = NULL;
905 	int error, n = ncpu, i;
906 	struct cpu_info *ci;
907 	CPU_INFO_ITERATOR cii;
908 
909 	/*
910 	 * if you specifically pass a buffer that is the size of the
911 	 * sum, or if you are probing for the size, you get the "sum"
912 	 * of cp_time (and the size thereof) across all processors.
913 	 *
914 	 * alternately, you can pass an additional mib number and get
915 	 * cp_time for that particular processor.
916 	 */
917 	switch (namelen) {
918 	case 0:
919 		if (*oldlenp == sizeof(uint64_t) * CPUSTATES || oldp == NULL) {
920 			node.sysctl_size = sizeof(uint64_t) * CPUSTATES;
921 			n = -1; /* SUM */
922 		}
923 		else {
924 			node.sysctl_size = n * sizeof(uint64_t) * CPUSTATES;
925 			n = -2; /* ALL */
926 		}
927 		break;
928 	case 1:
929 		if (name[0] < 0 || name[0] >= n)
930 			return (ENOENT); /* ENOSUCHPROCESSOR */
931 		node.sysctl_size = sizeof(uint64_t) * CPUSTATES;
932 		n = name[0];
933 		/*
934 		 * adjust these so that sysctl_lookup() will be happy
935 		 */
936 		name++;
937 		namelen--;
938 		break;
939 	default:
940 		return (EINVAL);
941 	}
942 
943 	cp_time = kmem_alloc(node.sysctl_size, KM_SLEEP);
944 	if (cp_time == NULL)
945 		return (ENOMEM);
946 	node.sysctl_data = cp_time;
947 	memset(cp_time, 0, node.sysctl_size);
948 
949 	for (CPU_INFO_FOREACH(cii, ci)) {
950 		if (n <= 0) {
951 			for (i = 0; i < CPUSTATES; i++) {
952 				cp_time[i] += ci->ci_schedstate.spc_cp_time[i];
953 			}
954 		}
955 		/*
956 		 * if a specific processor was requested and we just
957 		 * did it, we're done here
958 		 */
959 		if (n == 0)
960 			break;
961 		/*
962 		 * if doing "all", skip to next cp_time set for next processor
963 		 */
964 		if (n == -2)
965 			cp_time += CPUSTATES;
966 		/*
967 		 * if we're doing a specific processor, we're one
968 		 * processor closer
969 		 */
970 		if (n > 0)
971 			n--;
972 	}
973 
974 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
975 	kmem_free(node.sysctl_data, node.sysctl_size);
976 	return (error);
977 }
978 
979 #if NPTY > 0
980 /*
981  * sysctl helper routine for kern.maxptys. Ensures that any new value
982  * is acceptable to the pty subsystem.
983  */
984 static int
985 sysctl_kern_maxptys(SYSCTLFN_ARGS)
986 {
987 	int pty_maxptys(int, int);		/* defined in kern/tty_pty.c */
988 	int error, xmax;
989 	struct sysctlnode node;
990 
991 	/* get current value of maxptys */
992 	xmax = pty_maxptys(0, 0);
993 
994 	node = *rnode;
995 	node.sysctl_data = &xmax;
996 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
997 	if (error || newp == NULL)
998 		return (error);
999 
1000 	if (xmax != pty_maxptys(xmax, 1))
1001 		return (EINVAL);
1002 
1003 	return (0);
1004 }
1005 #endif /* NPTY > 0 */
1006 
1007 /*
1008  * sysctl helper routine to do kern.lwp.* work.
1009  */
1010 static int
1011 sysctl_kern_lwp(SYSCTLFN_ARGS)
1012 {
1013 	struct kinfo_lwp klwp;
1014 	struct proc *p;
1015 	struct lwp *l2, *l3;
1016 	char *where, *dp;
1017 	int pid, elem_size, elem_count;
1018 	int buflen, needed, error;
1019 	bool gotit;
1020 
1021 	if (namelen == 1 && name[0] == CTL_QUERY)
1022 		return (sysctl_query(SYSCTLFN_CALL(rnode)));
1023 
1024 	dp = where = oldp;
1025 	buflen = where != NULL ? *oldlenp : 0;
1026 	error = needed = 0;
1027 
1028 	if (newp != NULL || namelen != 3)
1029 		return (EINVAL);
1030 	pid = name[0];
1031 	elem_size = name[1];
1032 	elem_count = name[2];
1033 
1034 	sysctl_unlock();
1035 	if (pid == -1) {
1036 		mutex_enter(proc_lock);
1037 		PROCLIST_FOREACH(p, &allproc) {
1038 			/* Grab a hold on the process. */
1039 			if (!rw_tryenter(&p->p_reflock, RW_READER)) {
1040 				continue;
1041 			}
1042 			mutex_exit(proc_lock);
1043 
1044 			mutex_enter(p->p_lock);
1045 			LIST_FOREACH(l2, &p->p_lwps, l_sibling) {
1046 				if (buflen >= elem_size && elem_count > 0) {
1047 					lwp_lock(l2);
1048 					fill_lwp(l2, &klwp);
1049 					lwp_unlock(l2);
1050 					mutex_exit(p->p_lock);
1051 
1052 					/*
1053 					 * Copy out elem_size, but not
1054 					 * larger than the size of a
1055 					 * struct kinfo_proc2.
1056 					 */
1057 					error = dcopyout(l, &klwp, dp,
1058 					    min(sizeof(klwp), elem_size));
1059 					if (error) {
1060 						rw_exit(&p->p_reflock);
1061 						goto cleanup;
1062 					}
1063 					mutex_enter(p->p_lock);
1064 					LIST_FOREACH(l3, &p->p_lwps,
1065 					    l_sibling) {
1066 						if (l2 == l3)
1067 							break;
1068 					}
1069 					if (l3 == NULL) {
1070 						mutex_exit(p->p_lock);
1071 						rw_exit(&p->p_reflock);
1072 						error = EAGAIN;
1073 						goto cleanup;
1074 					}
1075 					dp += elem_size;
1076 					buflen -= elem_size;
1077 					elem_count--;
1078 				}
1079 				needed += elem_size;
1080 			}
1081 			mutex_exit(p->p_lock);
1082 
1083 			/* Drop reference to process. */
1084 			mutex_enter(proc_lock);
1085 			rw_exit(&p->p_reflock);
1086 		}
1087 		mutex_exit(proc_lock);
1088 	} else {
1089 		mutex_enter(proc_lock);
1090 		p = proc_find(pid);
1091 		if (p == NULL) {
1092 			error = ESRCH;
1093 			mutex_exit(proc_lock);
1094 			goto cleanup;
1095 		}
1096 		/* Grab a hold on the process. */
1097 		gotit = rw_tryenter(&p->p_reflock, RW_READER);
1098 		mutex_exit(proc_lock);
1099 		if (!gotit) {
1100 			error = ESRCH;
1101 			goto cleanup;
1102 		}
1103 
1104 		mutex_enter(p->p_lock);
1105 		LIST_FOREACH(l2, &p->p_lwps, l_sibling) {
1106 			if (buflen >= elem_size && elem_count > 0) {
1107 				lwp_lock(l2);
1108 				fill_lwp(l2, &klwp);
1109 				lwp_unlock(l2);
1110 				mutex_exit(p->p_lock);
1111 				/*
1112 				 * Copy out elem_size, but not larger than
1113 				 * the size of a struct kinfo_proc2.
1114 				 */
1115 				error = dcopyout(l, &klwp, dp,
1116 				    min(sizeof(klwp), elem_size));
1117 				if (error) {
1118 					rw_exit(&p->p_reflock);
1119 					goto cleanup;
1120 				}
1121 				mutex_enter(p->p_lock);
1122 				LIST_FOREACH(l3, &p->p_lwps, 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 		rw_exit(&p->p_reflock);
1142 	}
1143 
1144 	if (where != NULL) {
1145 		*oldlenp = dp - where;
1146 		if (needed > *oldlenp) {
1147 			sysctl_relock();
1148 			return (ENOMEM);
1149 		}
1150 	} else {
1151 		needed += KERN_LWPSLOP;
1152 		*oldlenp = needed;
1153 	}
1154 	error = 0;
1155  cleanup:
1156 	sysctl_relock();
1157 	return (error);
1158 }
1159 
1160 /*
1161  * sysctl helper routine for kern.forkfsleep node. Ensures that the
1162  * given value is not too large or two small, and is at least one
1163  * timer tick if not zero.
1164  */
1165 static int
1166 sysctl_kern_forkfsleep(SYSCTLFN_ARGS)
1167 {
1168 	/* userland sees value in ms, internally is in ticks */
1169 	extern int forkfsleep;		/* defined in kern/kern_fork.c */
1170 	int error, timo, lsleep;
1171 	struct sysctlnode node;
1172 
1173 	lsleep = forkfsleep * 1000 / hz;
1174 	node = *rnode;
1175 	node.sysctl_data = &lsleep;
1176 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1177 	if (error || newp == NULL)
1178 		return (error);
1179 
1180 	/* refuse negative values, and overly 'long time' */
1181 	if (lsleep < 0 || lsleep > MAXSLP * 1000)
1182 		return (EINVAL);
1183 
1184 	timo = mstohz(lsleep);
1185 
1186 	/* if the interval is >0 ms && <1 tick, use 1 tick */
1187 	if (lsleep != 0 && timo == 0)
1188 		forkfsleep = 1;
1189 	else
1190 		forkfsleep = timo;
1191 
1192 	return (0);
1193 }
1194 
1195 /*
1196  * sysctl helper routine for kern.root_partition
1197  */
1198 static int
1199 sysctl_kern_root_partition(SYSCTLFN_ARGS)
1200 {
1201 	int rootpart = DISKPART(rootdev);
1202 	struct sysctlnode node = *rnode;
1203 
1204 	node.sysctl_data = &rootpart;
1205 	return (sysctl_lookup(SYSCTLFN_CALL(&node)));
1206 }
1207 
1208 /*
1209  * sysctl helper function for kern.drivers
1210  */
1211 static int
1212 sysctl_kern_drivers(SYSCTLFN_ARGS)
1213 {
1214 	int error;
1215 	size_t buflen;
1216 	struct kinfo_drivers kd;
1217 	char *start, *where;
1218 	const char *dname;
1219 	int i;
1220 	extern struct devsw_conv *devsw_conv;
1221 	extern int max_devsw_convs;
1222 
1223 	start = where = oldp;
1224 	buflen = *oldlenp;
1225 	if (where == NULL) {
1226 		*oldlenp = max_devsw_convs * sizeof kd;
1227 		return 0;
1228 	}
1229 
1230 	/*
1231 	 * An array of kinfo_drivers structures
1232 	 */
1233 	error = 0;
1234 	sysctl_unlock();
1235 	mutex_enter(&device_lock);
1236 	for (i = 0; i < max_devsw_convs; i++) {
1237 		dname = devsw_conv[i].d_name;
1238 		if (dname == NULL)
1239 			continue;
1240 		if (buflen < sizeof kd) {
1241 			error = ENOMEM;
1242 			break;
1243 		}
1244 		memset(&kd, 0, sizeof(kd));
1245 		kd.d_bmajor = devsw_conv[i].d_bmajor;
1246 		kd.d_cmajor = devsw_conv[i].d_cmajor;
1247 		strlcpy(kd.d_name, dname, sizeof kd.d_name);
1248 		mutex_exit(&device_lock);
1249 		error = dcopyout(l, &kd, where, sizeof kd);
1250 		mutex_enter(&device_lock);
1251 		if (error != 0)
1252 			break;
1253 		buflen -= sizeof kd;
1254 		where += sizeof kd;
1255 	}
1256 	mutex_exit(&device_lock);
1257 	sysctl_relock();
1258 	*oldlenp = where - start;
1259 	return error;
1260 }
1261 
1262 static int
1263 sysctl_security_setidcore(SYSCTLFN_ARGS)
1264 {
1265 	int newsize, error;
1266 	struct sysctlnode node;
1267 
1268 	node = *rnode;
1269 	node.sysctl_data = &newsize;
1270 	newsize = *(int *)rnode->sysctl_data;
1271 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1272 	if (error || newp == NULL)
1273 		return error;
1274 
1275 	if (kauth_authorize_system(l->l_cred, KAUTH_SYSTEM_SETIDCORE,
1276 	    0, NULL, NULL, NULL))
1277 		return (EPERM);
1278 
1279 	*(int *)rnode->sysctl_data = newsize;
1280 
1281 	return 0;
1282 }
1283 
1284 static int
1285 sysctl_security_setidcorename(SYSCTLFN_ARGS)
1286 {
1287 	int error;
1288 	char *newsetidcorename;
1289 	struct sysctlnode node;
1290 
1291 	newsetidcorename = PNBUF_GET();
1292 	node = *rnode;
1293 	node.sysctl_data = newsetidcorename;
1294 	memcpy(node.sysctl_data, rnode->sysctl_data, MAXPATHLEN);
1295 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1296 	if (error || newp == NULL) {
1297 		goto out;
1298 	}
1299 	if (kauth_authorize_system(l->l_cred, KAUTH_SYSTEM_SETIDCORE,
1300 	    0, NULL, NULL, NULL)) {
1301 		error = EPERM;
1302 		goto out;
1303 	}
1304 	if (strlen(newsetidcorename) == 0) {
1305 		error = EINVAL;
1306 		goto out;
1307 	}
1308 	memcpy(rnode->sysctl_data, node.sysctl_data, MAXPATHLEN);
1309 out:
1310 	PNBUF_PUT(newsetidcorename);
1311 	return error;
1312 }
1313 
1314 /*
1315  * sysctl helper routine for kern.cp_id node. Maps cpus to their
1316  * cpuids.
1317  */
1318 static int
1319 sysctl_kern_cpid(SYSCTLFN_ARGS)
1320 {
1321 	struct sysctlnode node = *rnode;
1322 	uint64_t *cp_id = NULL;
1323 	int error, n = ncpu;
1324 	struct cpu_info *ci;
1325 	CPU_INFO_ITERATOR cii;
1326 
1327 	/*
1328 	 * Here you may either retrieve a single cpu id or the whole
1329 	 * set. The size you get back when probing depends on what
1330 	 * you ask for.
1331 	 */
1332 	switch (namelen) {
1333 	case 0:
1334 		node.sysctl_size = n * sizeof(uint64_t);
1335 		n = -2; /* ALL */
1336 		break;
1337 	case 1:
1338 		if (name[0] < 0 || name[0] >= n)
1339 			return (ENOENT); /* ENOSUCHPROCESSOR */
1340 		node.sysctl_size = sizeof(uint64_t);
1341 		n = name[0];
1342 		/*
1343 		 * adjust these so that sysctl_lookup() will be happy
1344 		 */
1345 		name++;
1346 		namelen--;
1347 		break;
1348 	default:
1349 		return (EINVAL);
1350 	}
1351 
1352 	cp_id = kmem_alloc(node.sysctl_size, KM_SLEEP);
1353 	if (cp_id == NULL)
1354 		return (ENOMEM);
1355 	node.sysctl_data = cp_id;
1356 	memset(cp_id, 0, node.sysctl_size);
1357 
1358 	for (CPU_INFO_FOREACH(cii, ci)) {
1359 		if (n <= 0)
1360 			cp_id[0] = cpu_index(ci);
1361 		/*
1362 		 * if a specific processor was requested and we just
1363 		 * did it, we're done here
1364 		 */
1365 		if (n == 0)
1366 			break;
1367 		/*
1368 		 * if doing "all", skip to next cp_id slot for next processor
1369 		 */
1370 		if (n == -2)
1371 			cp_id++;
1372 		/*
1373 		 * if we're doing a specific processor, we're one
1374 		 * processor closer
1375 		 */
1376 		if (n > 0)
1377 			n--;
1378 	}
1379 
1380 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1381 	kmem_free(node.sysctl_data, node.sysctl_size);
1382 	return (error);
1383 }
1384 
1385 /*
1386  * sysctl helper routine for hw.usermem and hw.usermem64. Values are
1387  * calculate on the fly taking into account integer overflow and the
1388  * current wired count.
1389  */
1390 static int
1391 sysctl_hw_usermem(SYSCTLFN_ARGS)
1392 {
1393 	u_int ui;
1394 	u_quad_t uq;
1395 	struct sysctlnode node;
1396 
1397 	node = *rnode;
1398 	switch (rnode->sysctl_num) {
1399 	case HW_USERMEM:
1400 		if ((ui = physmem - uvmexp.wired) > (UINT_MAX / PAGE_SIZE))
1401 			ui = UINT_MAX;
1402 		else
1403 			ui *= PAGE_SIZE;
1404 		node.sysctl_data = &ui;
1405 		break;
1406 	case HW_USERMEM64:
1407 		uq = (u_quad_t)(physmem - uvmexp.wired) * PAGE_SIZE;
1408 		node.sysctl_data = &uq;
1409 		break;
1410 	default:
1411 		return (EINVAL);
1412 	}
1413 
1414 	return (sysctl_lookup(SYSCTLFN_CALL(&node)));
1415 }
1416 
1417 /*
1418  * sysctl helper routine for kern.cnmagic node. Pulls the old value
1419  * out, encoded, and stuffs the new value in for decoding.
1420  */
1421 static int
1422 sysctl_hw_cnmagic(SYSCTLFN_ARGS)
1423 {
1424 	char magic[CNS_LEN];
1425 	int error;
1426 	struct sysctlnode node;
1427 
1428 	if (oldp)
1429 		cn_get_magic(magic, CNS_LEN);
1430 	node = *rnode;
1431 	node.sysctl_data = &magic[0];
1432 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1433 	if (error || newp == NULL)
1434 		return (error);
1435 
1436 	return (cn_set_magic(magic));
1437 }
1438 
1439 /*
1440  * ********************************************************************
1441  * section 3: public helper routines that are used for more than one
1442  * node
1443  * ********************************************************************
1444  */
1445 
1446 /*
1447  * sysctl helper routine for the kern.root_device node and some ports'
1448  * machdep.root_device nodes.
1449  */
1450 int
1451 sysctl_root_device(SYSCTLFN_ARGS)
1452 {
1453 	struct sysctlnode node;
1454 
1455 	node = *rnode;
1456 	node.sysctl_data = __UNCONST(device_xname(root_device));
1457 	node.sysctl_size = strlen(device_xname(root_device)) + 1;
1458 	return (sysctl_lookup(SYSCTLFN_CALL(&node)));
1459 }
1460 
1461 /*
1462  * sysctl helper routine for kern.consdev, dependent on the current
1463  * state of the console. Also used for machdep.console_device on some
1464  * ports.
1465  */
1466 int
1467 sysctl_consdev(SYSCTLFN_ARGS)
1468 {
1469 	dev_t consdev;
1470 	uint32_t oconsdev;
1471 	struct sysctlnode node;
1472 
1473 	if (cn_tab != NULL)
1474 		consdev = cn_tab->cn_dev;
1475 	else
1476 		consdev = NODEV;
1477 	node = *rnode;
1478 	switch (*oldlenp) {
1479 	case sizeof(consdev):
1480 		node.sysctl_data = &consdev;
1481 		node.sysctl_size = sizeof(consdev);
1482 		break;
1483 	case sizeof(oconsdev):
1484 		oconsdev = (uint32_t)consdev;
1485 		node.sysctl_data = &oconsdev;
1486 		node.sysctl_size = sizeof(oconsdev);
1487 		break;
1488 	default:
1489 		return EINVAL;
1490 	}
1491 	return (sysctl_lookup(SYSCTLFN_CALL(&node)));
1492 }
1493 
1494 /*
1495  * ********************************************************************
1496  * section 4: support for some helpers
1497  * ********************************************************************
1498  */
1499 
1500 
1501 /*
1502  * Fill in a kinfo_lwp structure for the specified lwp.
1503  */
1504 static void
1505 fill_lwp(struct lwp *l, struct kinfo_lwp *kl)
1506 {
1507 	struct proc *p = l->l_proc;
1508 	struct timeval tv;
1509 
1510 	KASSERT(lwp_locked(l, NULL));
1511 
1512 	memset(kl, 0, sizeof(*kl));
1513 
1514 	kl->l_forw = 0;
1515 	kl->l_back = 0;
1516 	kl->l_laddr = PTRTOUINT64(l);
1517 	kl->l_addr = PTRTOUINT64(l->l_addr);
1518 	kl->l_stat = l->l_stat;
1519 	kl->l_lid = l->l_lid;
1520 	kl->l_flag = L_INMEM;
1521 	kl->l_flag |= sysctl_map_flags(sysctl_lwpprflagmap, l->l_prflag);
1522 	kl->l_flag |= sysctl_map_flags(sysctl_lwpflagmap, l->l_flag);
1523 
1524 	kl->l_swtime = l->l_swtime;
1525 	kl->l_slptime = l->l_slptime;
1526 	if (l->l_stat == LSONPROC)
1527 		kl->l_schedflags = l->l_cpu->ci_schedstate.spc_flags;
1528 	else
1529 		kl->l_schedflags = 0;
1530 	kl->l_priority = lwp_eprio(l);
1531 	kl->l_usrpri = l->l_priority;
1532 	if (l->l_wchan)
1533 		strncpy(kl->l_wmesg, l->l_wmesg, sizeof(kl->l_wmesg));
1534 	kl->l_wchan = PTRTOUINT64(l->l_wchan);
1535 	kl->l_cpuid = cpu_index(l->l_cpu);
1536 	bintime2timeval(&l->l_rtime, &tv);
1537 	kl->l_rtime_sec = tv.tv_sec;
1538 	kl->l_rtime_usec = tv.tv_usec;
1539 	kl->l_cpticks = l->l_cpticks;
1540 	kl->l_pctcpu = l->l_pctcpu;
1541 	kl->l_pid = p->p_pid;
1542 	if (l->l_name == NULL)
1543 		kl->l_name[0] = '\0';
1544 	else
1545 		strlcpy(kl->l_name, l->l_name, sizeof(kl->l_name));
1546 }
1547