xref: /netbsd-src/sys/kern/init_sysctl.c (revision 154bfe8e089c1a0a4e9ed8414f08d3da90949162)
1 /*	$NetBSD: init_sysctl.c,v 1.226 2020/05/23 23:42:43 ad 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 1.226 2020/05/23 23:42:43 ad Exp $");
34 
35 #include "opt_sysv.h"
36 #include "opt_compat_netbsd.h"
37 #include "opt_modular.h"
38 #include "opt_gprof.h"
39 #include "pty.h"
40 
41 #include <sys/types.h>
42 #include <sys/param.h>
43 #include <sys/sysctl.h>
44 #include <sys/cpu.h>
45 #include <sys/errno.h>
46 #include <sys/systm.h>
47 #include <sys/kernel.h>
48 #include <sys/unistd.h>
49 #include <sys/disklabel.h>
50 #include <sys/cprng.h>
51 #include <sys/vnode_impl.h>	/* For vfs_drainvnodes(). */
52 #include <sys/mount.h>
53 #include <sys/namei.h>
54 #include <dev/cons.h>
55 #include <sys/socketvar.h>
56 #include <sys/file.h>
57 #include <sys/filedesc.h>
58 #include <sys/tty.h>
59 #include <sys/kmem.h>
60 #include <sys/reboot.h>
61 #include <sys/resource.h>
62 #include <sys/resourcevar.h>
63 #include <sys/exec.h>
64 #include <sys/conf.h>
65 #include <sys/device.h>
66 #include <sys/stat.h>
67 #include <sys/kauth.h>
68 #include <sys/ktrace.h>
69 
70 #include <sys/cpu.h>
71 
72 int security_setidcore_dump;
73 char security_setidcore_path[MAXPATHLEN] = "/var/crash/%n.core";
74 uid_t security_setidcore_owner = 0;
75 gid_t security_setidcore_group = 0;
76 mode_t security_setidcore_mode = (S_IRUSR|S_IWUSR);
77 
78 /*
79  * Current status of SysV IPC capability.  Initially, these are
80  * 0 if the capability is not built-in to the kernel, but can
81  * be updated if the appropriate kernel module is (auto)loaded.
82  */
83 
84 int kern_has_sysvmsg = 0;
85 int kern_has_sysvshm = 0;
86 int kern_has_sysvsem = 0;
87 
88 static const u_int sysctl_lwpprflagmap[] = {
89 	LPR_DETACHED, L_DETACHED,
90 	0
91 };
92 
93 /*
94  * try over estimating by 5 procs/lwps
95  */
96 #define KERN_LWPSLOP	(5 * sizeof(struct kinfo_lwp))
97 
98 static int dcopyout(struct lwp *, const void *, void *, size_t);
99 
100 static int
101 dcopyout(struct lwp *l, const void *kaddr, void *uaddr, size_t len)
102 {
103 	int error;
104 
105 	error = copyout(kaddr, uaddr, len);
106 	ktrmibio(-1, UIO_READ, uaddr, len, error);
107 
108 	return error;
109 }
110 
111 static int sysctl_kern_maxvnodes(SYSCTLFN_PROTO);
112 static int sysctl_kern_messages(SYSCTLFN_PROTO);
113 static int sysctl_kern_boottime(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 		       sysctl_kern_boottime, 0, NULL, sizeof(struct timespec),
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 	sysctl_createv(clog, 0, NULL, NULL,
501 		       CTLFLAG_PERMANENT,
502 		       CTLTYPE_INT, "root_partition",
503 		       SYSCTL_DESCR("Root partition on the root device"),
504 		       sysctl_kern_root_partition, 0, NULL, 0,
505 		       CTL_KERN, KERN_ROOT_PARTITION, CTL_EOL);
506 	sysctl_createv(clog, 0, NULL, NULL,
507 		       CTLFLAG_PERMANENT,
508 		       CTLTYPE_STRUCT, "drivers",
509 		       SYSCTL_DESCR("List of all drivers with block and "
510 				    "character device numbers"),
511 		       sysctl_kern_drivers, 0, NULL, 0,
512 		       CTL_KERN, KERN_DRIVERS, CTL_EOL);
513 	sysctl_createv(clog, 0, NULL, NULL,
514 		       CTLFLAG_PERMANENT,
515 		       CTLTYPE_STRUCT, "cp_id",
516 		       SYSCTL_DESCR("Mapping of CPU number to CPU id"),
517 		       sysctl_kern_cpid, 0, NULL, 0,
518 		       CTL_KERN, KERN_CP_ID, CTL_EOL);
519 	sysctl_createv(clog, 0, NULL, &rnode,
520 		       CTLFLAG_PERMANENT,
521 		       CTLTYPE_NODE, "coredump",
522 		       SYSCTL_DESCR("Coredump settings."),
523 		       NULL, 0, NULL, 0,
524 		       CTL_KERN, CTL_CREATE, CTL_EOL);
525 	sysctl_createv(clog, 0, &rnode, &rnode,
526 		       CTLFLAG_PERMANENT,
527 		       CTLTYPE_NODE, "setid",
528 		       SYSCTL_DESCR("Set-id processes' coredump settings."),
529 		       NULL, 0, NULL, 0,
530 		       CTL_CREATE, CTL_EOL);
531 	sysctl_createv(clog, 0, &rnode, NULL,
532 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
533 		       CTLTYPE_INT, "dump",
534 		       SYSCTL_DESCR("Allow set-id processes to dump core."),
535 		       sysctl_security_setidcore, 0, &security_setidcore_dump,
536 		       sizeof(security_setidcore_dump),
537 		       CTL_CREATE, CTL_EOL);
538 	sysctl_createv(clog, 0, &rnode, NULL,
539 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
540 		       CTLTYPE_STRING, "path",
541 		       SYSCTL_DESCR("Path pattern for set-id coredumps."),
542 		       sysctl_security_setidcorename, 0,
543 		       security_setidcore_path,
544 		       sizeof(security_setidcore_path),
545 		       CTL_CREATE, CTL_EOL);
546 	sysctl_createv(clog, 0, &rnode, NULL,
547 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
548 		       CTLTYPE_INT, "owner",
549 		       SYSCTL_DESCR("Owner id for set-id processes' cores."),
550 		       sysctl_security_setidcore, 0, &security_setidcore_owner,
551 		       0,
552 		       CTL_CREATE, CTL_EOL);
553 	sysctl_createv(clog, 0, &rnode, NULL,
554 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
555 		       CTLTYPE_INT, "group",
556 		       SYSCTL_DESCR("Group id for set-id processes' cores."),
557 		       sysctl_security_setidcore, 0, &security_setidcore_group,
558 		       0,
559 		       CTL_CREATE, CTL_EOL);
560 	sysctl_createv(clog, 0, &rnode, NULL,
561 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
562 		       CTLTYPE_INT, "mode",
563 		       SYSCTL_DESCR("Mode for set-id processes' cores."),
564 		       sysctl_security_setidcore, 0, &security_setidcore_mode,
565 		       0,
566 		       CTL_CREATE, CTL_EOL);
567 	sysctl_createv(clog, 0, NULL, NULL,
568 		       CTLFLAG_IMMEDIATE|CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
569 		       CTLTYPE_INT, "no_sa_support",
570 		       SYSCTL_DESCR("0 if the kernel supports SA, otherwise "
571 		       "it doesn't"),
572 		       NULL, 1, NULL, 0,
573 		       CTL_KERN, CTL_CREATE, CTL_EOL);
574 	sysctl_createv(clog, 0, NULL, NULL,
575 			CTLFLAG_PERMANENT,
576 			CTLTYPE_STRING, "configname",
577 			SYSCTL_DESCR("Name of config file"),
578 			NULL, 0, __UNCONST(kernel_ident), 0,
579 			CTL_KERN, CTL_CREATE, CTL_EOL);
580 	sysctl_createv(clog, 0, NULL, NULL,
581 			CTLFLAG_PERMANENT,
582 			CTLTYPE_STRING, "buildinfo",
583 			SYSCTL_DESCR("Information from build environment"),
584 			NULL, 0, __UNCONST(buildinfo), 0,
585 			CTL_KERN, CTL_CREATE, CTL_EOL);
586 	sysctl_createv(clog, 0, NULL, NULL,
587 			CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
588 			CTLTYPE_INT, "messages",
589 			SYSCTL_DESCR("Kernel message verbosity"),
590 			sysctl_kern_messages, 0, NULL, 0,
591 			CTL_KERN, CTL_CREATE, CTL_EOL);
592 }
593 
594 SYSCTL_SETUP(sysctl_hw_misc_setup, "sysctl hw subtree misc setup")
595 {
596 
597 	sysctl_createv(clog, 0, NULL, NULL,
598 		       CTLFLAG_PERMANENT,
599 		       CTLTYPE_INT, "usermem",
600 		       SYSCTL_DESCR("Bytes of non-kernel memory"),
601 		       sysctl_hw_usermem, 0, NULL, 0,
602 		       CTL_HW, HW_USERMEM, CTL_EOL);
603 	sysctl_createv(clog, 0, NULL, NULL,
604 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE|CTLFLAG_HEX,
605 		       CTLTYPE_STRING, "cnmagic",
606 		       SYSCTL_DESCR("Console magic key sequence"),
607 		       sysctl_hw_cnmagic, 0, NULL, CNS_LEN,
608 		       CTL_HW, HW_CNMAGIC, CTL_EOL);
609 	sysctl_createv(clog, 0, NULL, NULL,
610 		       CTLFLAG_PERMANENT,
611 		       CTLTYPE_QUAD, "usermem64",
612 		       SYSCTL_DESCR("Bytes of non-kernel memory"),
613 		       sysctl_hw_usermem, 0, NULL, 0,
614 		       CTL_HW, HW_USERMEM64, CTL_EOL);
615 }
616 
617 #ifdef DEBUG
618 /*
619  * Debugging related system variables.
620  */
621 struct ctldebug /* debug0, */ /* debug1, */ debug2, debug3, debug4;
622 struct ctldebug debug5, debug6, debug7, debug8, debug9;
623 struct ctldebug debug10, debug11, debug12, debug13, debug14;
624 struct ctldebug debug15, debug16, debug17, debug18, debug19;
625 static struct ctldebug *debugvars[] = {
626 	&debug0, &debug1, &debug2, &debug3, &debug4,
627 	&debug5, &debug6, &debug7, &debug8, &debug9,
628 	&debug10, &debug11, &debug12, &debug13, &debug14,
629 	&debug15, &debug16, &debug17, &debug18, &debug19,
630 };
631 
632 /*
633  * this setup routine is a replacement for debug_sysctl()
634  *
635  * note that it creates several nodes per defined debug variable
636  */
637 SYSCTL_SETUP(sysctl_debug_setup, "sysctl debug subtree setup")
638 {
639 	struct ctldebug *cdp;
640 	char nodename[20];
641 	int i;
642 
643 	/*
644 	 * two ways here:
645 	 *
646 	 * the "old" way (debug.name -> value) which was emulated by
647 	 * the sysctl(8) binary
648 	 *
649 	 * the new way, which the sysctl(8) binary was actually using
650 
651 	 node	debug
652 	 node	debug.0
653 	 string debug.0.name
654 	 int	debug.0.value
655 	 int	debug.name
656 
657 	 */
658 
659 	for (i = 0; i < __arraycount(debugvars); i++) {
660 		cdp = debugvars[i];
661 		if (cdp->debugname == NULL || cdp->debugvar == NULL)
662 			continue;
663 
664 		snprintf(nodename, sizeof(nodename), "debug%d", i);
665 		sysctl_createv(clog, 0, NULL, NULL,
666 			       CTLFLAG_PERMANENT|CTLFLAG_HIDDEN,
667 			       CTLTYPE_NODE, nodename, NULL,
668 			       NULL, 0, NULL, 0,
669 			       CTL_DEBUG, i, CTL_EOL);
670 		sysctl_createv(clog, 0, NULL, NULL,
671 			       CTLFLAG_PERMANENT|CTLFLAG_HIDDEN,
672 			       CTLTYPE_STRING, "name", NULL,
673 			       /*XXXUNCONST*/
674 			       NULL, 0, __UNCONST(cdp->debugname), 0,
675 			       CTL_DEBUG, i, CTL_DEBUG_NAME, CTL_EOL);
676 		sysctl_createv(clog, 0, NULL, NULL,
677 			       CTLFLAG_PERMANENT|CTLFLAG_HIDDEN,
678 			       CTLTYPE_INT, "value", NULL,
679 			       NULL, 0, cdp->debugvar, 0,
680 			       CTL_DEBUG, i, CTL_DEBUG_VALUE, CTL_EOL);
681 		sysctl_createv(clog, 0, NULL, NULL,
682 			       CTLFLAG_PERMANENT,
683 			       CTLTYPE_INT, cdp->debugname, NULL,
684 			       NULL, 0, cdp->debugvar, 0,
685 			       CTL_DEBUG, CTL_CREATE, CTL_EOL);
686 	}
687 }
688 #endif /* DEBUG */
689 
690 /*
691  * ********************************************************************
692  * section 2: private node-specific helper routines.
693  * ********************************************************************
694  */
695 
696 /*
697  * sysctl helper routine for kern.maxvnodes.  Drain vnodes if
698  * new value is lower than desiredvnodes and then calls reinit
699  * routines that needs to adjust to the new value.
700  */
701 static int
702 sysctl_kern_maxvnodes(SYSCTLFN_ARGS)
703 {
704 	int error, new_vnodes, old_vnodes, new_max;
705 	struct sysctlnode node;
706 
707 	new_vnodes = desiredvnodes;
708 	node = *rnode;
709 	node.sysctl_data = &new_vnodes;
710 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
711 	if (error || newp == NULL)
712 		return (error);
713 
714 	/*
715 	 * sysctl passes down unsigned values, require them
716 	 * to be positive
717 	 */
718 	if (new_vnodes <= 0)
719 		return (EINVAL);
720 
721 	/* Limits: 75% of kmem and physical memory. */
722 	new_max = calc_cache_size(vmem_size(kmem_arena, VMEM_FREE|VMEM_ALLOC),
723 	    75, 75) / VNODE_COST;
724 	if (new_vnodes > new_max)
725 		new_vnodes = new_max;
726 
727 	old_vnodes = desiredvnodes;
728 	desiredvnodes = new_vnodes;
729 	error = vfs_drainvnodes();
730 	if (error) {
731 		desiredvnodes = old_vnodes;
732 		return (error);
733 	}
734 	vfs_reinit();
735 
736 	return (0);
737 }
738 
739 /*
740  * sysctl helper routine for kern.messages.
741  * Alters boothowto to display kernel messages in increasing verbosity
742  * from 0 to 4.
743  */
744 
745 #define MAXMESSAGES            4
746 static int
747 sysctl_kern_messages(SYSCTLFN_ARGS)
748 {
749 	int error, messageverbose, messagemask, newboothowto;
750 	struct sysctlnode node;
751 
752 	messagemask = (AB_NORMAL|AB_QUIET|AB_SILENT|AB_VERBOSE|AB_DEBUG);
753 	switch (boothowto & messagemask) {
754 	case AB_SILENT:
755 		messageverbose = 0;
756 		break;
757 	case AB_QUIET:
758 		messageverbose = 1;
759 		break;
760 	case AB_VERBOSE:
761 		messageverbose = 3;
762 		break;
763 	case AB_DEBUG:
764 		messageverbose = 4;
765 		break;
766 	case AB_NORMAL:
767 	default:
768 		messageverbose = 2;
769 	}
770 
771 	node = *rnode;
772 	node.sysctl_data = &messageverbose;
773 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
774 	if (error || newp == NULL)
775 		return (error);
776 	if (messageverbose < 0 || messageverbose > MAXMESSAGES)
777 		return EINVAL;
778 
779 	/* Set boothowto */
780 	newboothowto = boothowto & ~messagemask;
781 
782 	switch (messageverbose) {
783 	case 0:
784 		newboothowto |= AB_SILENT;
785 		break;
786 	case 1:
787 		newboothowto |= AB_QUIET;
788 		break;
789 	case 3:
790 		newboothowto |= AB_VERBOSE;
791 		break;
792 	case 4:
793 		newboothowto |= AB_DEBUG;
794 		break;
795 	case 2:
796 	default:                /* Messages default to normal. */
797 		break;
798 	}
799 
800 	boothowto = newboothowto;
801 
802 	return (0);
803 }
804 
805 /*
806  * sysctl helper routine for the kern.boottime node
807  */
808 static int
809 sysctl_kern_boottime(SYSCTLFN_ARGS)
810 {
811 	struct sysctlnode node;
812 	struct timespec ts;
813 
814 	getnanoboottime(&ts);
815 	node = *rnode;
816 	node.sysctl_data = &ts;
817 	return (sysctl_lookup(SYSCTLFN_CALL(&node)));
818 }
819 
820 /*
821  * sysctl helper routine for rtc_offset - set time after changes
822  */
823 static int
824 sysctl_kern_rtc_offset(SYSCTLFN_ARGS)
825 {
826 	struct timespec ts, delta;
827 	int error, new_rtc_offset;
828 	struct sysctlnode node;
829 
830 	new_rtc_offset = rtc_offset;
831 	node = *rnode;
832 	node.sysctl_data = &new_rtc_offset;
833 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
834 	if (error || newp == NULL)
835 		return (error);
836 
837 	if (kauth_authorize_system(l->l_cred, KAUTH_SYSTEM_TIME,
838 	    KAUTH_REQ_SYSTEM_TIME_RTCOFFSET,
839 	    KAUTH_ARG(new_rtc_offset), NULL, NULL))
840 		return (EPERM);
841 	if (rtc_offset == new_rtc_offset)
842 		return (0);
843 
844 	/* if we change the offset, adjust the time */
845 	nanotime(&ts);
846 	delta.tv_sec = 60 * (new_rtc_offset - rtc_offset);
847 	delta.tv_nsec = 0;
848 	timespecadd(&ts, &delta, &ts);
849 	rtc_offset = new_rtc_offset;
850 	return (settime(l->l_proc, &ts));
851 }
852 
853 /*
854  * sysctl helper routine for kern.maxproc. Ensures that the new
855  * values are not too low or too high.
856  */
857 static int
858 sysctl_kern_maxproc(SYSCTLFN_ARGS)
859 {
860 	int error, nmaxproc;
861 	struct sysctlnode node;
862 
863 	nmaxproc = maxproc;
864 	node = *rnode;
865 	node.sysctl_data = &nmaxproc;
866 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
867 	if (error || newp == NULL)
868 		return (error);
869 
870 	if (nmaxproc < 0 || nmaxproc >= PID_MAX)
871 		return (EINVAL);
872 #ifdef __HAVE_CPU_MAXPROC
873 	if (nmaxproc > cpu_maxproc())
874 		return (EINVAL);
875 #endif
876 	error = 0;
877 #ifdef __HAVE_MAXPROC_HOOK
878 	error = cpu_maxproc_hook(nmaxproc);
879 #endif
880 	if (error)
881 		return error;
882 
883 	maxproc = nmaxproc;
884 
885 	return (0);
886 }
887 
888 /*
889  * sysctl helper function for kern.hostid. The hostid is a long, but
890  * we export it as an int, so we need to give it a little help.
891  */
892 static int
893 sysctl_kern_hostid(SYSCTLFN_ARGS)
894 {
895 	int error, inthostid;
896 	struct sysctlnode node;
897 
898 	inthostid = hostid;  /* XXX assumes sizeof int <= sizeof long */
899 	node = *rnode;
900 	node.sysctl_data = &inthostid;
901 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
902 	if (error || newp == NULL)
903 		return (error);
904 
905 	hostid = (unsigned)inthostid;
906 
907 	return (0);
908 }
909 
910 /*
911  * sysctl helper routine for kern.defcorename. In the case of a new
912  * string being assigned, check that it's not a zero-length string.
913  * (XXX the check in -current doesn't work, but do we really care?)
914  */
915 static int
916 sysctl_kern_defcorename(SYSCTLFN_ARGS)
917 {
918 	int error;
919 	char *newcorename;
920 	struct sysctlnode node;
921 
922 	newcorename = PNBUF_GET();
923 	node = *rnode;
924 	node.sysctl_data = &newcorename[0];
925 	memcpy(node.sysctl_data, rnode->sysctl_data, MAXPATHLEN);
926 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
927 	if (error || newp == NULL) {
928 		goto done;
929 	}
930 
931 	/*
932 	 * when sysctl_lookup() deals with a string, it's guaranteed
933 	 * to come back nul terminated. So there.  :)
934 	 */
935 	if (strlen(newcorename) == 0) {
936 		error = EINVAL;
937 	} else {
938 		memcpy(rnode->sysctl_data, node.sysctl_data, MAXPATHLEN);
939 		error = 0;
940 	}
941 done:
942 	PNBUF_PUT(newcorename);
943 	return error;
944 }
945 
946 /*
947  * sysctl helper routine for kern.cp_time node. Adds up cpu time
948  * across all cpus.
949  */
950 static int
951 sysctl_kern_cptime(SYSCTLFN_ARGS)
952 {
953 	struct sysctlnode node = *rnode;
954 	uint64_t *cp_time = NULL;
955 	int error, n = ncpu, i;
956 	struct cpu_info *ci;
957 	CPU_INFO_ITERATOR cii;
958 
959 	/*
960 	 * if you specifically pass a buffer that is the size of the
961 	 * sum, or if you are probing for the size, you get the "sum"
962 	 * of cp_time (and the size thereof) across all processors.
963 	 *
964 	 * alternately, you can pass an additional mib number and get
965 	 * cp_time for that particular processor.
966 	 */
967 	switch (namelen) {
968 	case 0:
969 		if (*oldlenp == sizeof(uint64_t) * CPUSTATES || oldp == NULL) {
970 			node.sysctl_size = sizeof(uint64_t) * CPUSTATES;
971 			n = -1; /* SUM */
972 		}
973 		else {
974 			node.sysctl_size = n * sizeof(uint64_t) * CPUSTATES;
975 			n = -2; /* ALL */
976 		}
977 		break;
978 	case 1:
979 		if (name[0] < 0 || name[0] >= n)
980 			return (ENOENT); /* ENOSUCHPROCESSOR */
981 		node.sysctl_size = sizeof(uint64_t) * CPUSTATES;
982 		n = name[0];
983 		/*
984 		 * adjust these so that sysctl_lookup() will be happy
985 		 */
986 		name++;
987 		namelen--;
988 		break;
989 	default:
990 		return (EINVAL);
991 	}
992 
993 	cp_time = kmem_alloc(node.sysctl_size, KM_SLEEP);
994 	node.sysctl_data = cp_time;
995 	memset(cp_time, 0, node.sysctl_size);
996 
997 	for (CPU_INFO_FOREACH(cii, ci)) {
998 		if (n <= 0) {
999 			for (i = 0; i < CPUSTATES; i++) {
1000 				cp_time[i] += ci->ci_schedstate.spc_cp_time[i];
1001 			}
1002 		}
1003 		/*
1004 		 * if a specific processor was requested and we just
1005 		 * did it, we're done here
1006 		 */
1007 		if (n == 0)
1008 			break;
1009 		/*
1010 		 * if doing "all", skip to next cp_time set for next processor
1011 		 */
1012 		if (n == -2)
1013 			cp_time += CPUSTATES;
1014 		/*
1015 		 * if we're doing a specific processor, we're one
1016 		 * processor closer
1017 		 */
1018 		if (n > 0)
1019 			n--;
1020 	}
1021 
1022 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1023 	kmem_free(node.sysctl_data, node.sysctl_size);
1024 	return (error);
1025 }
1026 
1027 #if NPTY > 0
1028 /*
1029  * sysctl helper routine for kern.maxptys. Ensures that any new value
1030  * is acceptable to the pty subsystem.
1031  */
1032 static int
1033 sysctl_kern_maxptys(SYSCTLFN_ARGS)
1034 {
1035 	int pty_maxptys(int, int);		/* defined in kern/tty_pty.c */
1036 	int error, xmax;
1037 	struct sysctlnode node;
1038 
1039 	/* get current value of maxptys */
1040 	xmax = pty_maxptys(0, 0);
1041 
1042 	node = *rnode;
1043 	node.sysctl_data = &xmax;
1044 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1045 	if (error || newp == NULL)
1046 		return (error);
1047 
1048 	if (xmax != pty_maxptys(xmax, 1))
1049 		return (EINVAL);
1050 
1051 	return (0);
1052 }
1053 #endif /* NPTY > 0 */
1054 
1055 /*
1056  * sysctl helper routine to do kern.lwp.* work.
1057  */
1058 static int
1059 sysctl_kern_lwp(SYSCTLFN_ARGS)
1060 {
1061 	struct kinfo_lwp klwp;
1062 	struct proc *p;
1063 	struct lwp *l2, *l3;
1064 	char *where, *dp;
1065 	int pid, elem_size, elem_count;
1066 	int buflen, needed, error;
1067 	bool gotit;
1068 
1069 	if (namelen == 1 && name[0] == CTL_QUERY)
1070 		return (sysctl_query(SYSCTLFN_CALL(rnode)));
1071 
1072 	dp = where = oldp;
1073 	buflen = where != NULL ? *oldlenp : 0;
1074 	error = needed = 0;
1075 
1076 	if (newp != NULL || namelen != 3)
1077 		return (EINVAL);
1078 	pid = name[0];
1079 	elem_size = name[1];
1080 	elem_count = name[2];
1081 
1082 	sysctl_unlock();
1083 	if (pid == -1) {
1084 		mutex_enter(&proc_lock);
1085 		PROCLIST_FOREACH(p, &allproc) {
1086 			/* Grab a hold on the process. */
1087 			if (!rw_tryenter(&p->p_reflock, RW_READER)) {
1088 				continue;
1089 			}
1090 			mutex_exit(&proc_lock);
1091 
1092 			mutex_enter(p->p_lock);
1093 			LIST_FOREACH(l2, &p->p_lwps, l_sibling) {
1094 				if (buflen >= elem_size && elem_count > 0) {
1095 					lwp_lock(l2);
1096 					fill_lwp(l2, &klwp);
1097 					lwp_unlock(l2);
1098 					mutex_exit(p->p_lock);
1099 
1100 					/*
1101 					 * Copy out elem_size, but not
1102 					 * larger than the size of a
1103 					 * struct kinfo_proc2.
1104 					 */
1105 					error = dcopyout(l, &klwp, dp,
1106 					    uimin(sizeof(klwp), elem_size));
1107 					if (error) {
1108 						rw_exit(&p->p_reflock);
1109 						goto cleanup;
1110 					}
1111 					mutex_enter(p->p_lock);
1112 					LIST_FOREACH(l3, &p->p_lwps,
1113 					    l_sibling) {
1114 						if (l2 == l3)
1115 							break;
1116 					}
1117 					if (l3 == NULL) {
1118 						mutex_exit(p->p_lock);
1119 						rw_exit(&p->p_reflock);
1120 						error = EAGAIN;
1121 						goto cleanup;
1122 					}
1123 					dp += elem_size;
1124 					buflen -= elem_size;
1125 					elem_count--;
1126 				}
1127 				needed += elem_size;
1128 			}
1129 			mutex_exit(p->p_lock);
1130 
1131 			/* Drop reference to process. */
1132 			mutex_enter(&proc_lock);
1133 			rw_exit(&p->p_reflock);
1134 		}
1135 		mutex_exit(&proc_lock);
1136 	} else {
1137 		mutex_enter(&proc_lock);
1138 		p = proc_find(pid);
1139 		if (p == NULL) {
1140 			error = ESRCH;
1141 			mutex_exit(&proc_lock);
1142 			goto cleanup;
1143 		}
1144 		/* Grab a hold on the process. */
1145 		gotit = rw_tryenter(&p->p_reflock, RW_READER);
1146 		mutex_exit(&proc_lock);
1147 		if (!gotit) {
1148 			error = ESRCH;
1149 			goto cleanup;
1150 		}
1151 
1152 		mutex_enter(p->p_lock);
1153 		LIST_FOREACH(l2, &p->p_lwps, l_sibling) {
1154 			if (buflen >= elem_size && elem_count > 0) {
1155 				lwp_lock(l2);
1156 				fill_lwp(l2, &klwp);
1157 				lwp_unlock(l2);
1158 				mutex_exit(p->p_lock);
1159 				/*
1160 				 * Copy out elem_size, but not larger than
1161 				 * the size of a struct kinfo_proc2.
1162 				 */
1163 				error = dcopyout(l, &klwp, dp,
1164 				    uimin(sizeof(klwp), elem_size));
1165 				if (error) {
1166 					rw_exit(&p->p_reflock);
1167 					goto cleanup;
1168 				}
1169 				mutex_enter(p->p_lock);
1170 				LIST_FOREACH(l3, &p->p_lwps, l_sibling) {
1171 					if (l2 == l3)
1172 						break;
1173 				}
1174 				if (l3 == NULL) {
1175 					mutex_exit(p->p_lock);
1176 					rw_exit(&p->p_reflock);
1177 					error = EAGAIN;
1178 					goto cleanup;
1179 				}
1180 				dp += elem_size;
1181 				buflen -= elem_size;
1182 				elem_count--;
1183 			}
1184 			needed += elem_size;
1185 		}
1186 		mutex_exit(p->p_lock);
1187 
1188 		/* Drop reference to process. */
1189 		rw_exit(&p->p_reflock);
1190 	}
1191 
1192 	if (where != NULL) {
1193 		*oldlenp = dp - where;
1194 		if (needed > *oldlenp) {
1195 			sysctl_relock();
1196 			return (ENOMEM);
1197 		}
1198 	} else {
1199 		needed += KERN_LWPSLOP;
1200 		*oldlenp = needed;
1201 	}
1202 	error = 0;
1203  cleanup:
1204 	sysctl_relock();
1205 	return (error);
1206 }
1207 
1208 /*
1209  * sysctl helper routine for kern.forkfsleep node. Ensures that the
1210  * given value is not too large or two small, and is at least one
1211  * timer tick if not zero.
1212  */
1213 static int
1214 sysctl_kern_forkfsleep(SYSCTLFN_ARGS)
1215 {
1216 	/* userland sees value in ms, internally is in ticks */
1217 	extern int forkfsleep;		/* defined in kern/kern_fork.c */
1218 	int error, timo, lsleep;
1219 	struct sysctlnode node;
1220 
1221 	lsleep = forkfsleep * 1000 / hz;
1222 	node = *rnode;
1223 	node.sysctl_data = &lsleep;
1224 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1225 	if (error || newp == NULL)
1226 		return (error);
1227 
1228 	/* refuse negative values, and overly 'long time' */
1229 	if (lsleep < 0 || lsleep > MAXSLP * 1000)
1230 		return (EINVAL);
1231 
1232 	timo = mstohz(lsleep);
1233 
1234 	/* if the interval is >0 ms && <1 tick, use 1 tick */
1235 	if (lsleep != 0 && timo == 0)
1236 		forkfsleep = 1;
1237 	else
1238 		forkfsleep = timo;
1239 
1240 	return (0);
1241 }
1242 
1243 /*
1244  * sysctl helper routine for kern.root_partition
1245  */
1246 static int
1247 sysctl_kern_root_partition(SYSCTLFN_ARGS)
1248 {
1249 	int rootpart = DISKPART(rootdev);
1250 	struct sysctlnode node = *rnode;
1251 
1252 	node.sysctl_data = &rootpart;
1253 	return (sysctl_lookup(SYSCTLFN_CALL(&node)));
1254 }
1255 
1256 /*
1257  * sysctl helper function for kern.drivers
1258  */
1259 static int
1260 sysctl_kern_drivers(SYSCTLFN_ARGS)
1261 {
1262 	int error;
1263 	size_t buflen;
1264 	struct kinfo_drivers kd;
1265 	char *start, *where;
1266 	const char *dname;
1267 	int i;
1268 	extern struct devsw_conv *devsw_conv;
1269 	extern int max_devsw_convs;
1270 
1271 	start = where = oldp;
1272 	buflen = *oldlenp;
1273 	if (where == NULL) {
1274 		*oldlenp = max_devsw_convs * sizeof kd;
1275 		return 0;
1276 	}
1277 
1278 	/*
1279 	 * An array of kinfo_drivers structures
1280 	 */
1281 	error = 0;
1282 	sysctl_unlock();
1283 	mutex_enter(&device_lock);
1284 	for (i = 0; i < max_devsw_convs; i++) {
1285 		dname = devsw_conv[i].d_name;
1286 		if (dname == NULL)
1287 			continue;
1288 		if (buflen < sizeof kd) {
1289 			error = ENOMEM;
1290 			break;
1291 		}
1292 		memset(&kd, 0, sizeof(kd));
1293 		kd.d_bmajor = devsw_conv[i].d_bmajor;
1294 		kd.d_cmajor = devsw_conv[i].d_cmajor;
1295 		strlcpy(kd.d_name, dname, sizeof kd.d_name);
1296 		mutex_exit(&device_lock);
1297 		error = dcopyout(l, &kd, where, sizeof kd);
1298 		mutex_enter(&device_lock);
1299 		if (error != 0)
1300 			break;
1301 		buflen -= sizeof kd;
1302 		where += sizeof kd;
1303 	}
1304 	mutex_exit(&device_lock);
1305 	sysctl_relock();
1306 	*oldlenp = where - start;
1307 	return error;
1308 }
1309 
1310 static int
1311 sysctl_security_setidcore(SYSCTLFN_ARGS)
1312 {
1313 	int newsize, error;
1314 	struct sysctlnode node;
1315 
1316 	node = *rnode;
1317 	node.sysctl_data = &newsize;
1318 	newsize = *(int *)rnode->sysctl_data;
1319 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1320 	if (error || newp == NULL)
1321 		return error;
1322 
1323 	if (kauth_authorize_system(l->l_cred, KAUTH_SYSTEM_SETIDCORE,
1324 	    0, NULL, NULL, NULL))
1325 		return (EPERM);
1326 
1327 	*(int *)rnode->sysctl_data = newsize;
1328 
1329 	return 0;
1330 }
1331 
1332 static int
1333 sysctl_security_setidcorename(SYSCTLFN_ARGS)
1334 {
1335 	int error;
1336 	char *newsetidcorename;
1337 	struct sysctlnode node;
1338 
1339 	newsetidcorename = PNBUF_GET();
1340 	node = *rnode;
1341 	node.sysctl_data = newsetidcorename;
1342 	memcpy(node.sysctl_data, rnode->sysctl_data, MAXPATHLEN);
1343 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1344 	if (error || newp == NULL) {
1345 		goto out;
1346 	}
1347 	if (kauth_authorize_system(l->l_cred, KAUTH_SYSTEM_SETIDCORE,
1348 	    0, NULL, NULL, NULL)) {
1349 		error = EPERM;
1350 		goto out;
1351 	}
1352 	if (strlen(newsetidcorename) == 0) {
1353 		error = EINVAL;
1354 		goto out;
1355 	}
1356 	memcpy(rnode->sysctl_data, node.sysctl_data, MAXPATHLEN);
1357 out:
1358 	PNBUF_PUT(newsetidcorename);
1359 	return error;
1360 }
1361 
1362 /*
1363  * sysctl helper routine for kern.cp_id node. Maps cpus to their
1364  * cpuids.
1365  */
1366 static int
1367 sysctl_kern_cpid(SYSCTLFN_ARGS)
1368 {
1369 	struct sysctlnode node = *rnode;
1370 	uint64_t *cp_id = NULL;
1371 	int error, n = ncpu;
1372 	struct cpu_info *ci;
1373 	CPU_INFO_ITERATOR cii;
1374 
1375 	/*
1376 	 * Here you may either retrieve a single cpu id or the whole
1377 	 * set. The size you get back when probing depends on what
1378 	 * you ask for.
1379 	 */
1380 	switch (namelen) {
1381 	case 0:
1382 		node.sysctl_size = n * sizeof(uint64_t);
1383 		n = -2; /* ALL */
1384 		break;
1385 	case 1:
1386 		if (name[0] < 0 || name[0] >= n)
1387 			return (ENOENT); /* ENOSUCHPROCESSOR */
1388 		node.sysctl_size = sizeof(uint64_t);
1389 		n = name[0];
1390 		/*
1391 		 * adjust these so that sysctl_lookup() will be happy
1392 		 */
1393 		name++;
1394 		namelen--;
1395 		break;
1396 	default:
1397 		return (EINVAL);
1398 	}
1399 
1400 	cp_id = kmem_alloc(node.sysctl_size, KM_SLEEP);
1401 	node.sysctl_data = cp_id;
1402 	memset(cp_id, 0, node.sysctl_size);
1403 
1404 	for (CPU_INFO_FOREACH(cii, ci)) {
1405 		if (n <= 0)
1406 			cp_id[0] = cpu_index(ci);
1407 		/*
1408 		 * if a specific processor was requested and we just
1409 		 * did it, we're done here
1410 		 */
1411 		if (n == 0)
1412 			break;
1413 		/*
1414 		 * if doing "all", skip to next cp_id slot for next processor
1415 		 */
1416 		if (n == -2)
1417 			cp_id++;
1418 		/*
1419 		 * if we're doing a specific processor, we're one
1420 		 * processor closer
1421 		 */
1422 		if (n > 0)
1423 			n--;
1424 	}
1425 
1426 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1427 	kmem_free(node.sysctl_data, node.sysctl_size);
1428 	return (error);
1429 }
1430 
1431 /*
1432  * sysctl helper routine for hw.usermem and hw.usermem64. Values are
1433  * calculate on the fly taking into account integer overflow and the
1434  * current wired count.
1435  */
1436 static int
1437 sysctl_hw_usermem(SYSCTLFN_ARGS)
1438 {
1439 	u_int ui;
1440 	u_quad_t uq;
1441 	struct sysctlnode node;
1442 
1443 	node = *rnode;
1444 	switch (rnode->sysctl_num) {
1445 	case HW_USERMEM:
1446 		if ((ui = physmem - uvmexp.wired) > (UINT_MAX / PAGE_SIZE))
1447 			ui = UINT_MAX;
1448 		else
1449 			ui *= PAGE_SIZE;
1450 		node.sysctl_data = &ui;
1451 		break;
1452 	case HW_USERMEM64:
1453 		uq = (u_quad_t)(physmem - uvmexp.wired) * PAGE_SIZE;
1454 		node.sysctl_data = &uq;
1455 		break;
1456 	default:
1457 		return (EINVAL);
1458 	}
1459 
1460 	return (sysctl_lookup(SYSCTLFN_CALL(&node)));
1461 }
1462 
1463 /*
1464  * sysctl helper routine for kern.cnmagic node. Pulls the old value
1465  * out, encoded, and stuffs the new value in for decoding.
1466  */
1467 static int
1468 sysctl_hw_cnmagic(SYSCTLFN_ARGS)
1469 {
1470 	char magic[CNS_LEN];
1471 	int error;
1472 	struct sysctlnode node;
1473 
1474 	if (oldp)
1475 		cn_get_magic(magic, CNS_LEN);
1476 	node = *rnode;
1477 	node.sysctl_data = &magic[0];
1478 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1479 	if (error || newp == NULL)
1480 		return (error);
1481 
1482 	return (cn_set_magic(magic));
1483 }
1484 
1485 /*
1486  * ********************************************************************
1487  * section 3: public helper routines that are used for more than one
1488  * node
1489  * ********************************************************************
1490  */
1491 
1492 /*
1493  * sysctl helper routine for the kern.root_device node and some ports'
1494  * machdep.root_device nodes.
1495  */
1496 int
1497 sysctl_root_device(SYSCTLFN_ARGS)
1498 {
1499 	struct sysctlnode node;
1500 
1501 	node = *rnode;
1502 	node.sysctl_data = __UNCONST(device_xname(root_device));
1503 	node.sysctl_size = strlen(device_xname(root_device)) + 1;
1504 	return (sysctl_lookup(SYSCTLFN_CALL(&node)));
1505 }
1506 
1507 /*
1508  * sysctl helper routine for kern.consdev, dependent on the current
1509  * state of the console. Also used for machdep.console_device on some
1510  * ports.
1511  */
1512 int
1513 sysctl_consdev(SYSCTLFN_ARGS)
1514 {
1515 	dev_t consdev;
1516 	uint32_t oconsdev;
1517 	struct sysctlnode node;
1518 
1519 	if (cn_tab != NULL)
1520 		consdev = cn_tab->cn_dev;
1521 	else
1522 		consdev = NODEV;
1523 	node = *rnode;
1524 	switch (*oldlenp) {
1525 	case sizeof(consdev):
1526 		node.sysctl_data = &consdev;
1527 		node.sysctl_size = sizeof(consdev);
1528 		break;
1529 	case sizeof(oconsdev):
1530 		oconsdev = (uint32_t)consdev;
1531 		node.sysctl_data = &oconsdev;
1532 		node.sysctl_size = sizeof(oconsdev);
1533 		break;
1534 	default:
1535 		return EINVAL;
1536 	}
1537 	return (sysctl_lookup(SYSCTLFN_CALL(&node)));
1538 }
1539 
1540 /*
1541  * ********************************************************************
1542  * section 4: support for some helpers
1543  * ********************************************************************
1544  */
1545 
1546 
1547 /*
1548  * Fill in a kinfo_lwp structure for the specified lwp.
1549  */
1550 static void
1551 fill_lwp(struct lwp *l, struct kinfo_lwp *kl)
1552 {
1553 	const bool allowaddr = get_expose_address(curproc);
1554 	struct proc *p = l->l_proc;
1555 	struct timeval tv;
1556 
1557 	KASSERT(lwp_locked(l, NULL));
1558 
1559 	memset(kl, 0, sizeof(*kl));
1560 
1561 	kl->l_forw = 0;
1562 	kl->l_back = 0;
1563 	COND_SET_VALUE(kl->l_laddr, PTRTOUINT64(l), allowaddr);
1564 	COND_SET_VALUE(kl->l_addr, PTRTOUINT64(l->l_addr), allowaddr);
1565 	kl->l_stat = l->l_stat;
1566 	kl->l_lid = l->l_lid;
1567 	kl->l_flag = L_INMEM;
1568 	kl->l_flag |= sysctl_map_flags(sysctl_lwpprflagmap, l->l_prflag);
1569 	kl->l_flag |= sysctl_map_flags(sysctl_lwpflagmap, l->l_flag);
1570 
1571 	kl->l_swtime = l->l_swtime;
1572 	kl->l_slptime = l->l_slptime;
1573 	if (l->l_stat == LSONPROC)
1574 		kl->l_schedflags = l->l_cpu->ci_schedstate.spc_flags;
1575 	else
1576 		kl->l_schedflags = 0;
1577 	kl->l_priority = lwp_eprio(l);
1578 	kl->l_usrpri = l->l_priority;
1579 	if (l->l_wchan)
1580 		strncpy(kl->l_wmesg, l->l_wmesg, sizeof(kl->l_wmesg));
1581 	COND_SET_VALUE(kl->l_wchan, PTRTOUINT64(l->l_wchan), allowaddr);
1582 	kl->l_cpuid = cpu_index(l->l_cpu);
1583 	bintime2timeval(&l->l_rtime, &tv);
1584 	kl->l_rtime_sec = tv.tv_sec;
1585 	kl->l_rtime_usec = tv.tv_usec;
1586 	kl->l_cpticks = l->l_cpticks;
1587 	kl->l_pctcpu = l->l_pctcpu;
1588 	kl->l_pid = p->p_pid;
1589 	if (l->l_name == NULL)
1590 		kl->l_name[0] = '\0';
1591 	else
1592 		strlcpy(kl->l_name, l->l_name, sizeof(kl->l_name));
1593 }
1594