xref: /netbsd-src/sys/kern/init_sysctl.c (revision bdc22b2e01993381dcefeff2bc9b56ca75a4235c)
1 /*	$NetBSD: init_sysctl.c,v 1.214 2018/02/04 17:31:51 maxv 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 1.214 2018/02/04 17:31:51 maxv 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 #ifdef DIAGNOSTIC
112 static int sysctl_kern_trigger_panic(SYSCTLFN_PROTO);
113 #endif
114 static int sysctl_kern_maxvnodes(SYSCTLFN_PROTO);
115 static int sysctl_kern_messages(SYSCTLFN_PROTO);
116 static int sysctl_kern_rtc_offset(SYSCTLFN_PROTO);
117 static int sysctl_kern_maxproc(SYSCTLFN_PROTO);
118 static int sysctl_kern_hostid(SYSCTLFN_PROTO);
119 static int sysctl_kern_defcorename(SYSCTLFN_PROTO);
120 static int sysctl_kern_cptime(SYSCTLFN_PROTO);
121 #if NPTY > 0
122 static int sysctl_kern_maxptys(SYSCTLFN_PROTO);
123 #endif /* NPTY > 0 */
124 static int sysctl_kern_lwp(SYSCTLFN_PROTO);
125 static int sysctl_kern_forkfsleep(SYSCTLFN_PROTO);
126 static int sysctl_kern_root_partition(SYSCTLFN_PROTO);
127 static int sysctl_kern_drivers(SYSCTLFN_PROTO);
128 static int sysctl_security_setidcore(SYSCTLFN_PROTO);
129 static int sysctl_security_setidcorename(SYSCTLFN_PROTO);
130 static int sysctl_kern_cpid(SYSCTLFN_PROTO);
131 static int sysctl_hw_usermem(SYSCTLFN_PROTO);
132 static int sysctl_hw_cnmagic(SYSCTLFN_PROTO);
133 
134 static void fill_lwp(struct lwp *l, struct kinfo_lwp *kl);
135 
136 /*
137  * ********************************************************************
138  * section 1: setup routines
139  * ********************************************************************
140  * These functions are stuffed into a link set for sysctl setup
141  * functions. They're never called or referenced from anywhere else.
142  * ********************************************************************
143  */
144 
145 /*
146  * this setup routine is a replacement for kern_sysctl()
147  */
148 SYSCTL_SETUP(sysctl_kern_setup, "sysctl kern subtree setup")
149 {
150 	extern int kern_logsigexit;	/* defined in kern/kern_sig.c */
151 	extern fixpt_t ccpu;		/* defined in kern/kern_synch.c */
152 	extern int dumponpanic;		/* defined in kern/subr_prf.c */
153 	const struct sysctlnode *rnode;
154 
155 	sysctl_createv(clog, 0, NULL, NULL,
156 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
157 		       CTLTYPE_INT, "maxvnodes",
158 		       SYSCTL_DESCR("Maximum number of vnodes"),
159 		       sysctl_kern_maxvnodes, 0, NULL, 0,
160 		       CTL_KERN, KERN_MAXVNODES, CTL_EOL);
161 	sysctl_createv(clog, 0, NULL, NULL,
162 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
163 		       CTLTYPE_INT, "maxproc",
164 		       SYSCTL_DESCR("Maximum number of simultaneous processes"),
165 		       sysctl_kern_maxproc, 0, NULL, 0,
166 		       CTL_KERN, KERN_MAXPROC, CTL_EOL);
167 	sysctl_createv(clog, 0, NULL, NULL,
168 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
169 		       CTLTYPE_INT, "maxfiles",
170 		       SYSCTL_DESCR("Maximum number of open files"),
171 		       NULL, 0, &maxfiles, 0,
172 		       CTL_KERN, KERN_MAXFILES, CTL_EOL);
173 	sysctl_createv(clog, 0, NULL, NULL,
174 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
175 		       CTLTYPE_INT, "argmax",
176 		       SYSCTL_DESCR("Maximum number of bytes of arguments to "
177 				    "execve(2)"),
178 		       NULL, ARG_MAX, NULL, 0,
179 		       CTL_KERN, KERN_ARGMAX, CTL_EOL);
180 	sysctl_createv(clog, 0, NULL, NULL,
181 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE|CTLFLAG_HEX,
182 		       CTLTYPE_INT, "hostid",
183 		       SYSCTL_DESCR("System host ID number"),
184 		       sysctl_kern_hostid, 0, NULL, 0,
185 		       CTL_KERN, KERN_HOSTID, CTL_EOL);
186 	sysctl_createv(clog, 0, NULL, NULL,
187 		       CTLFLAG_PERMANENT,
188 		       CTLTYPE_STRUCT, "vnode",
189 		       SYSCTL_DESCR("System vnode table"),
190 		       sysctl_kern_vnode, 0, NULL, 0,
191 		       CTL_KERN, KERN_VNODE, CTL_EOL);
192 #ifndef GPROF
193 	sysctl_createv(clog, 0, NULL, NULL,
194 		       CTLFLAG_PERMANENT,
195 		       CTLTYPE_NODE, "profiling",
196 		       SYSCTL_DESCR("Profiling information (not available)"),
197 		       sysctl_notavail, 0, NULL, 0,
198 		       CTL_KERN, KERN_PROF, CTL_EOL);
199 #endif
200 	sysctl_createv(clog, 0, NULL, NULL,
201 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
202 		       CTLTYPE_INT, "posix1version",
203 		       SYSCTL_DESCR("Version of ISO/IEC 9945 (POSIX 1003.1) "
204 				    "with which the operating system attempts "
205 				    "to comply"),
206 		       NULL, _POSIX_VERSION, NULL, 0,
207 		       CTL_KERN, KERN_POSIX1, CTL_EOL);
208 	sysctl_createv(clog, 0, NULL, NULL,
209 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
210 		       CTLTYPE_INT, "ngroups",
211 		       SYSCTL_DESCR("Maximum number of supplemental groups"),
212 		       NULL, NGROUPS_MAX, NULL, 0,
213 		       CTL_KERN, KERN_NGROUPS, CTL_EOL);
214 	sysctl_createv(clog, 0, NULL, NULL,
215 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
216 		       CTLTYPE_INT, "job_control",
217 		       SYSCTL_DESCR("Whether job control is available"),
218 		       NULL, 1, NULL, 0,
219 		       CTL_KERN, KERN_JOB_CONTROL, CTL_EOL);
220 	sysctl_createv(clog, 0, NULL, NULL,
221 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
222 		       CTLTYPE_INT, "saved_ids",
223 		       SYSCTL_DESCR("Whether POSIX saved set-group/user ID is "
224 				    "available"), NULL,
225 #ifdef _POSIX_SAVED_IDS
226 		       1,
227 #else /* _POSIX_SAVED_IDS */
228 		       0,
229 #endif /* _POSIX_SAVED_IDS */
230 		       NULL, 0, CTL_KERN, KERN_SAVED_IDS, CTL_EOL);
231 	sysctl_createv(clog, 0, NULL, NULL,
232 		       CTLFLAG_PERMANENT|CTLFLAG_HEX,
233 		       CTLTYPE_INT, "boothowto",
234 		       SYSCTL_DESCR("Flags from boot loader"),
235 		       NULL, 0, &boothowto, sizeof(boothowto),
236 		       CTL_KERN, CTL_CREATE, CTL_EOL);
237 	sysctl_createv(clog, 0, NULL, NULL,
238 		       CTLFLAG_PERMANENT,
239 		       CTLTYPE_STRUCT, "boottime",
240 		       SYSCTL_DESCR("System boot time"),
241 		       NULL, 0, &boottime, sizeof(boottime),
242 		       CTL_KERN, KERN_BOOTTIME, CTL_EOL);
243 	sysctl_createv(clog, 0, NULL, NULL,
244 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
245 		       CTLTYPE_INT, "maxpartitions",
246 		       SYSCTL_DESCR("Maximum number of partitions allowed per "
247 				    "disk"),
248 		       NULL, MAXPARTITIONS, NULL, 0,
249 		       CTL_KERN, KERN_MAXPARTITIONS, CTL_EOL);
250 	sysctl_createv(clog, 0, NULL, NULL,
251 		       CTLFLAG_PERMANENT,
252 		       CTLTYPE_STRUCT, "timex", NULL,
253 		       sysctl_notavail, 0, NULL, 0,
254 		       CTL_KERN, KERN_TIMEX, CTL_EOL);
255 	sysctl_createv(clog, 0, NULL, NULL,
256 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
257 		       CTLTYPE_INT, "rtc_offset",
258 		       SYSCTL_DESCR("Offset of real time clock from UTC in "
259 				    "minutes"),
260 		       sysctl_kern_rtc_offset, 0, &rtc_offset, 0,
261 		       CTL_KERN, KERN_RTC_OFFSET, CTL_EOL);
262 	sysctl_createv(clog, 0, NULL, NULL,
263 		       CTLFLAG_PERMANENT,
264 		       CTLTYPE_STRING, "root_device",
265 		       SYSCTL_DESCR("Name of the root device"),
266 		       sysctl_root_device, 0, NULL, 0,
267 		       CTL_KERN, KERN_ROOT_DEVICE, CTL_EOL);
268 	sysctl_createv(clog, 0, NULL, NULL,
269 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
270 		       CTLTYPE_INT, "fsync",
271 		       SYSCTL_DESCR("Whether the POSIX 1003.1b File "
272 				    "Synchronization Option is available on "
273 				    "this system"),
274 		       NULL, 1, NULL, 0,
275 		       CTL_KERN, KERN_FSYNC, CTL_EOL);
276 	sysctl_createv(clog, 0, NULL, NULL,
277 		       CTLFLAG_PERMANENT,
278 		       CTLTYPE_NODE, "ipc",
279 		       SYSCTL_DESCR("SysV IPC options"),
280 		       NULL, 0, NULL, 0,
281 		       CTL_KERN, KERN_SYSVIPC, CTL_EOL);
282 	sysctl_createv(clog, 0, NULL, NULL,
283 		       CTLFLAG_PERMANENT|CTLFLAG_READONLY,
284 		       CTLTYPE_INT, "sysvmsg",
285 		       SYSCTL_DESCR("System V style message support available"),
286 		       NULL, 0, &kern_has_sysvmsg, sizeof(int),
287 		       CTL_KERN, KERN_SYSVIPC, KERN_SYSVIPC_MSG, CTL_EOL);
288 	sysctl_createv(clog, 0, NULL, NULL,
289 		       CTLFLAG_PERMANENT|CTLFLAG_READONLY,
290 		       CTLTYPE_INT, "sysvsem",
291 		       SYSCTL_DESCR("System V style semaphore support "
292 				    "available"),
293 		       NULL, 0, &kern_has_sysvsem, sizeof(int),
294 		       CTL_KERN, KERN_SYSVIPC, KERN_SYSVIPC_SEM, CTL_EOL);
295 	sysctl_createv(clog, 0, NULL, NULL,
296 		       CTLFLAG_PERMANENT|CTLFLAG_READONLY,
297 		       CTLTYPE_INT, "sysvshm",
298 		       SYSCTL_DESCR("System V style shared memory support "
299 				    "available"),
300 		       NULL, 0, &kern_has_sysvshm, sizeof(int),
301 		       CTL_KERN, KERN_SYSVIPC, KERN_SYSVIPC_SHM, CTL_EOL);
302 	sysctl_createv(clog, 0, NULL, NULL,
303 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
304 		       CTLTYPE_INT, "synchronized_io",
305 		       SYSCTL_DESCR("Whether the POSIX 1003.1b Synchronized "
306 				    "I/O Option is available on this system"),
307 		       NULL, 1, NULL, 0,
308 		       CTL_KERN, KERN_SYNCHRONIZED_IO, CTL_EOL);
309 	sysctl_createv(clog, 0, NULL, NULL,
310 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
311 		       CTLTYPE_INT, "iov_max",
312 		       SYSCTL_DESCR("Maximum number of iovec structures per "
313 				    "process"),
314 		       NULL, IOV_MAX, NULL, 0,
315 		       CTL_KERN, KERN_IOV_MAX, CTL_EOL);
316 	sysctl_createv(clog, 0, NULL, NULL,
317 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
318 		       CTLTYPE_INT, "mapped_files",
319 		       SYSCTL_DESCR("Whether the POSIX 1003.1b Memory Mapped "
320 				    "Files Option is available on this system"),
321 		       NULL, 1, NULL, 0,
322 		       CTL_KERN, KERN_MAPPED_FILES, CTL_EOL);
323 	sysctl_createv(clog, 0, NULL, NULL,
324 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
325 		       CTLTYPE_INT, "memlock",
326 		       SYSCTL_DESCR("Whether the POSIX 1003.1b Process Memory "
327 				    "Locking Option is available on this "
328 				    "system"),
329 		       NULL, 1, NULL, 0,
330 		       CTL_KERN, KERN_MEMLOCK, CTL_EOL);
331 	sysctl_createv(clog, 0, NULL, NULL,
332 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
333 		       CTLTYPE_INT, "memlock_range",
334 		       SYSCTL_DESCR("Whether the POSIX 1003.1b Range Memory "
335 				    "Locking Option is available on this "
336 				    "system"),
337 		       NULL, 1, NULL, 0,
338 		       CTL_KERN, KERN_MEMLOCK_RANGE, CTL_EOL);
339 	sysctl_createv(clog, 0, NULL, NULL,
340 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
341 		       CTLTYPE_INT, "memory_protection",
342 		       SYSCTL_DESCR("Whether the POSIX 1003.1b Memory "
343 				    "Protection Option is available on this "
344 				    "system"),
345 		       NULL, 1, NULL, 0,
346 		       CTL_KERN, KERN_MEMORY_PROTECTION, CTL_EOL);
347 	sysctl_createv(clog, 0, NULL, NULL,
348 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
349 		       CTLTYPE_INT, "login_name_max",
350 		       SYSCTL_DESCR("Maximum login name length"),
351 		       NULL, LOGIN_NAME_MAX, NULL, 0,
352 		       CTL_KERN, KERN_LOGIN_NAME_MAX, CTL_EOL);
353 	sysctl_createv(clog, 0, NULL, NULL,
354 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
355 		       CTLTYPE_STRING, "defcorename",
356 		       SYSCTL_DESCR("Default core file name"),
357 		       sysctl_kern_defcorename, 0, defcorename, MAXPATHLEN,
358 		       CTL_KERN, KERN_DEFCORENAME, CTL_EOL);
359 	sysctl_createv(clog, 0, NULL, NULL,
360 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
361 		       CTLTYPE_INT, "logsigexit",
362 		       SYSCTL_DESCR("Log process exit when caused by signals"),
363 		       NULL, 0, &kern_logsigexit, 0,
364 		       CTL_KERN, KERN_LOGSIGEXIT, CTL_EOL);
365 	sysctl_createv(clog, 0, NULL, NULL,
366 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
367 		       CTLTYPE_INT, "fscale",
368 		       SYSCTL_DESCR("Kernel fixed-point scale factor"),
369 		       NULL, FSCALE, NULL, 0,
370 		       CTL_KERN, KERN_FSCALE, CTL_EOL);
371 	sysctl_createv(clog, 0, NULL, NULL,
372 		       CTLFLAG_PERMANENT,
373 		       CTLTYPE_INT, "ccpu",
374 		       SYSCTL_DESCR("Scheduler exponential decay value"),
375 		       NULL, 0, &ccpu, 0,
376 		       CTL_KERN, KERN_CCPU, CTL_EOL);
377 	sysctl_createv(clog, 0, NULL, NULL,
378 		       CTLFLAG_PERMANENT,
379 		       CTLTYPE_STRUCT, "cp_time",
380 		       SYSCTL_DESCR("Clock ticks spent in different CPU states"),
381 		       sysctl_kern_cptime, 0, NULL, 0,
382 		       CTL_KERN, KERN_CP_TIME, CTL_EOL);
383 	sysctl_createv(clog, 0, NULL, NULL,
384 		       CTLFLAG_PERMANENT,
385 		       CTLTYPE_STRUCT, "consdev",
386 		       SYSCTL_DESCR("Console device"),
387 		       sysctl_consdev, 0, NULL, sizeof(dev_t),
388 		       CTL_KERN, KERN_CONSDEV, CTL_EOL);
389 #if NPTY > 0
390 	sysctl_createv(clog, 0, NULL, NULL,
391 		       CTLFLAG_PERMANENT,
392 		       CTLTYPE_INT, "maxptys",
393 		       SYSCTL_DESCR("Maximum number of pseudo-ttys"),
394 		       sysctl_kern_maxptys, 0, NULL, 0,
395 		       CTL_KERN, KERN_MAXPTYS, CTL_EOL);
396 #endif /* NPTY > 0 */
397 	sysctl_createv(clog, 0, NULL, NULL,
398 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
399 		       CTLTYPE_INT, "maxphys",
400 		       SYSCTL_DESCR("Maximum raw I/O transfer size"),
401 		       NULL, MAXPHYS, NULL, 0,
402 		       CTL_KERN, KERN_MAXPHYS, CTL_EOL);
403 	sysctl_createv(clog, 0, NULL, NULL,
404 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
405 		       CTLTYPE_INT, "monotonic_clock",
406 		       SYSCTL_DESCR("Implementation version of the POSIX "
407 				    "1003.1b Monotonic Clock Option"),
408 		       /* XXX _POSIX_VERSION */
409 		       NULL, _POSIX_MONOTONIC_CLOCK, NULL, 0,
410 		       CTL_KERN, KERN_MONOTONIC_CLOCK, CTL_EOL);
411 	sysctl_createv(clog, 0, NULL, NULL,
412 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
413 		       CTLTYPE_INT, "labelsector",
414 		       SYSCTL_DESCR("Sector number containing the disklabel"),
415 		       NULL, LABELSECTOR, NULL, 0,
416 		       CTL_KERN, KERN_LABELSECTOR, CTL_EOL);
417 	sysctl_createv(clog, 0, NULL, NULL,
418 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
419 		       CTLTYPE_INT, "labeloffset",
420 		       SYSCTL_DESCR("Offset of the disklabel within the "
421 				    "sector"),
422 		       NULL, LABELOFFSET, NULL, 0,
423 		       CTL_KERN, KERN_LABELOFFSET, CTL_EOL);
424 	sysctl_createv(clog, 0, NULL, NULL,
425 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
426 		       CTLTYPE_INT, "labelusesmbr",
427 		       SYSCTL_DESCR("disklabel is inside MBR partition"),
428 		       NULL, LABELUSESMBR, NULL, 0,
429 		       CTL_KERN, CTL_CREATE, CTL_EOL);
430 	sysctl_createv(clog, 0, NULL, NULL,
431 		       CTLFLAG_PERMANENT,
432 		       CTLTYPE_NODE, "lwp",
433 		       SYSCTL_DESCR("System-wide LWP information"),
434 		       sysctl_kern_lwp, 0, NULL, 0,
435 		       CTL_KERN, KERN_LWP, CTL_EOL);
436 	sysctl_createv(clog, 0, NULL, NULL,
437 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
438 		       CTLTYPE_INT, "forkfsleep",
439 		       SYSCTL_DESCR("Milliseconds to sleep on fork failure due "
440 				    "to process limits"),
441 		       sysctl_kern_forkfsleep, 0, NULL, 0,
442 		       CTL_KERN, KERN_FORKFSLEEP, CTL_EOL);
443 	sysctl_createv(clog, 0, NULL, NULL,
444 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
445 		       CTLTYPE_INT, "posix_threads",
446 		       SYSCTL_DESCR("Version of IEEE Std 1003.1 and its "
447 				    "Threads option to which the system "
448 				    "attempts to conform"),
449 		       /* XXX _POSIX_VERSION */
450 		       NULL, _POSIX_THREADS, NULL, 0,
451 		       CTL_KERN, KERN_POSIX_THREADS, CTL_EOL);
452 	sysctl_createv(clog, 0, NULL, NULL,
453 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
454 		       CTLTYPE_INT, "posix_semaphores",
455 		       SYSCTL_DESCR("Version of IEEE Std 1003.1 and its "
456 				    "Semaphores option to which the system "
457 				    "attempts to conform"), NULL,
458 		       200112, NULL, 0,
459 		       CTL_KERN, KERN_POSIX_SEMAPHORES, CTL_EOL);
460 	sysctl_createv(clog, 0, NULL, NULL,
461 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
462 		       CTLTYPE_INT, "posix_barriers",
463 		       SYSCTL_DESCR("Version of IEEE Std 1003.1 and its "
464 				    "Barriers option to which the system "
465 				    "attempts to conform"),
466 		       /* XXX _POSIX_VERSION */
467 		       NULL, _POSIX_BARRIERS, NULL, 0,
468 		       CTL_KERN, KERN_POSIX_BARRIERS, CTL_EOL);
469 	sysctl_createv(clog, 0, NULL, NULL,
470 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
471 		       CTLTYPE_INT, "posix_timers",
472 		       SYSCTL_DESCR("Version of IEEE Std 1003.1 and its "
473 				    "Timers option to which the system "
474 				    "attempts to conform"),
475 		       /* XXX _POSIX_VERSION */
476 		       NULL, _POSIX_TIMERS, NULL, 0,
477 		       CTL_KERN, KERN_POSIX_TIMERS, CTL_EOL);
478 	sysctl_createv(clog, 0, NULL, NULL,
479 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
480 		       CTLTYPE_INT, "posix_spin_locks",
481 		       SYSCTL_DESCR("Version of IEEE Std 1003.1 and its Spin "
482 				    "Locks option to which the system attempts "
483 				    "to conform"),
484 		       /* XXX _POSIX_VERSION */
485 		       NULL, _POSIX_SPIN_LOCKS, NULL, 0,
486 		       CTL_KERN, KERN_POSIX_SPIN_LOCKS, CTL_EOL);
487 	sysctl_createv(clog, 0, NULL, NULL,
488 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
489 		       CTLTYPE_INT, "posix_reader_writer_locks",
490 		       SYSCTL_DESCR("Version of IEEE Std 1003.1 and its "
491 				    "Read-Write Locks option to which the "
492 				    "system attempts to conform"),
493 		       /* XXX _POSIX_VERSION */
494 		       NULL, _POSIX_READER_WRITER_LOCKS, NULL, 0,
495 		       CTL_KERN, KERN_POSIX_READER_WRITER_LOCKS, CTL_EOL);
496 	sysctl_createv(clog, 0, NULL, NULL,
497 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
498 		       CTLTYPE_INT, "dump_on_panic",
499 		       SYSCTL_DESCR("Perform a crash dump on system panic"),
500 		       NULL, 0, &dumponpanic, 0,
501 		       CTL_KERN, KERN_DUMP_ON_PANIC, CTL_EOL);
502 #ifdef DIAGNOSTIC
503 	sysctl_createv(clog, 0, NULL, NULL,
504 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
505 		       CTLTYPE_INT, "panic_now",
506 		       SYSCTL_DESCR("Trigger a panic"),
507 		       sysctl_kern_trigger_panic, 0, NULL, 0,
508 		       CTL_KERN, CTL_CREATE, CTL_EOL);
509 #endif
510 	sysctl_createv(clog, 0, NULL, NULL,
511 		       CTLFLAG_PERMANENT,
512 		       CTLTYPE_INT, "root_partition",
513 		       SYSCTL_DESCR("Root partition on the root device"),
514 		       sysctl_kern_root_partition, 0, NULL, 0,
515 		       CTL_KERN, KERN_ROOT_PARTITION, CTL_EOL);
516 	sysctl_createv(clog, 0, NULL, NULL,
517 		       CTLFLAG_PERMANENT,
518 		       CTLTYPE_STRUCT, "drivers",
519 		       SYSCTL_DESCR("List of all drivers with block and "
520 				    "character device numbers"),
521 		       sysctl_kern_drivers, 0, NULL, 0,
522 		       CTL_KERN, KERN_DRIVERS, CTL_EOL);
523 	sysctl_createv(clog, 0, NULL, NULL,
524 		       CTLFLAG_PERMANENT,
525 		       CTLTYPE_STRUCT, "cp_id",
526 		       SYSCTL_DESCR("Mapping of CPU number to CPU id"),
527 		       sysctl_kern_cpid, 0, NULL, 0,
528 		       CTL_KERN, KERN_CP_ID, CTL_EOL);
529 	sysctl_createv(clog, 0, NULL, &rnode,
530 		       CTLFLAG_PERMANENT,
531 		       CTLTYPE_NODE, "coredump",
532 		       SYSCTL_DESCR("Coredump settings."),
533 		       NULL, 0, NULL, 0,
534 		       CTL_KERN, CTL_CREATE, CTL_EOL);
535 	sysctl_createv(clog, 0, &rnode, &rnode,
536 		       CTLFLAG_PERMANENT,
537 		       CTLTYPE_NODE, "setid",
538 		       SYSCTL_DESCR("Set-id processes' coredump settings."),
539 		       NULL, 0, NULL, 0,
540 		       CTL_CREATE, CTL_EOL);
541 	sysctl_createv(clog, 0, &rnode, NULL,
542 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
543 		       CTLTYPE_INT, "dump",
544 		       SYSCTL_DESCR("Allow set-id processes to dump core."),
545 		       sysctl_security_setidcore, 0, &security_setidcore_dump,
546 		       sizeof(security_setidcore_dump),
547 		       CTL_CREATE, CTL_EOL);
548 	sysctl_createv(clog, 0, &rnode, NULL,
549 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
550 		       CTLTYPE_STRING, "path",
551 		       SYSCTL_DESCR("Path pattern for set-id coredumps."),
552 		       sysctl_security_setidcorename, 0,
553 		       security_setidcore_path,
554 		       sizeof(security_setidcore_path),
555 		       CTL_CREATE, CTL_EOL);
556 	sysctl_createv(clog, 0, &rnode, NULL,
557 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
558 		       CTLTYPE_INT, "owner",
559 		       SYSCTL_DESCR("Owner id for set-id processes' cores."),
560 		       sysctl_security_setidcore, 0, &security_setidcore_owner,
561 		       0,
562 		       CTL_CREATE, CTL_EOL);
563 	sysctl_createv(clog, 0, &rnode, NULL,
564 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
565 		       CTLTYPE_INT, "group",
566 		       SYSCTL_DESCR("Group id for set-id processes' cores."),
567 		       sysctl_security_setidcore, 0, &security_setidcore_group,
568 		       0,
569 		       CTL_CREATE, CTL_EOL);
570 	sysctl_createv(clog, 0, &rnode, NULL,
571 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
572 		       CTLTYPE_INT, "mode",
573 		       SYSCTL_DESCR("Mode for set-id processes' cores."),
574 		       sysctl_security_setidcore, 0, &security_setidcore_mode,
575 		       0,
576 		       CTL_CREATE, CTL_EOL);
577 	sysctl_createv(clog, 0, NULL, NULL,
578 		       CTLFLAG_IMMEDIATE|CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
579 		       CTLTYPE_INT, "no_sa_support",
580 		       SYSCTL_DESCR("0 if the kernel supports SA, otherwise "
581 		       "it doesn't"),
582 		       NULL, 1, NULL, 0,
583 		       CTL_KERN, CTL_CREATE, CTL_EOL);
584 	sysctl_createv(clog, 0, NULL, NULL,
585 			CTLFLAG_PERMANENT,
586 			CTLTYPE_STRING, "configname",
587 			SYSCTL_DESCR("Name of config file"),
588 			NULL, 0, __UNCONST(kernel_ident), 0,
589 			CTL_KERN, CTL_CREATE, CTL_EOL);
590 	sysctl_createv(clog, 0, NULL, NULL,
591 			CTLFLAG_PERMANENT,
592 			CTLTYPE_STRING, "buildinfo",
593 			SYSCTL_DESCR("Information from build environment"),
594 			NULL, 0, __UNCONST(buildinfo), 0,
595 			CTL_KERN, CTL_CREATE, CTL_EOL);
596 	sysctl_createv(clog, 0, NULL, NULL,
597 			CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
598 			CTLTYPE_INT, "messages",
599 			SYSCTL_DESCR("Kernel message verbosity"),
600 			sysctl_kern_messages, 0, NULL, 0,
601 			CTL_KERN, CTL_CREATE, CTL_EOL);
602 }
603 
604 SYSCTL_SETUP(sysctl_hw_misc_setup, "sysctl hw subtree misc setup")
605 {
606 
607 	sysctl_createv(clog, 0, NULL, NULL,
608 		       CTLFLAG_PERMANENT,
609 		       CTLTYPE_INT, "usermem",
610 		       SYSCTL_DESCR("Bytes of non-kernel memory"),
611 		       sysctl_hw_usermem, 0, NULL, 0,
612 		       CTL_HW, HW_USERMEM, CTL_EOL);
613 	sysctl_createv(clog, 0, NULL, NULL,
614 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE|CTLFLAG_HEX,
615 		       CTLTYPE_STRING, "cnmagic",
616 		       SYSCTL_DESCR("Console magic key sequence"),
617 		       sysctl_hw_cnmagic, 0, NULL, CNS_LEN,
618 		       CTL_HW, HW_CNMAGIC, CTL_EOL);
619 	sysctl_createv(clog, 0, NULL, NULL,
620 		       CTLFLAG_PERMANENT,
621 		       CTLTYPE_QUAD, "usermem64",
622 		       SYSCTL_DESCR("Bytes of non-kernel memory"),
623 		       sysctl_hw_usermem, 0, NULL, 0,
624 		       CTL_HW, HW_USERMEM64, CTL_EOL);
625 }
626 
627 #ifdef DEBUG
628 /*
629  * Debugging related system variables.
630  */
631 struct ctldebug /* debug0, */ /* debug1, */ debug2, debug3, debug4;
632 struct ctldebug debug5, debug6, debug7, debug8, debug9;
633 struct ctldebug debug10, debug11, debug12, debug13, debug14;
634 struct ctldebug debug15, debug16, debug17, debug18, debug19;
635 static struct ctldebug *debugvars[CTL_DEBUG_MAXID] = {
636 	&debug0, &debug1, &debug2, &debug3, &debug4,
637 	&debug5, &debug6, &debug7, &debug8, &debug9,
638 	&debug10, &debug11, &debug12, &debug13, &debug14,
639 	&debug15, &debug16, &debug17, &debug18, &debug19,
640 };
641 
642 /*
643  * this setup routine is a replacement for debug_sysctl()
644  *
645  * note that it creates several nodes per defined debug variable
646  */
647 SYSCTL_SETUP(sysctl_debug_setup, "sysctl debug subtree setup")
648 {
649 	struct ctldebug *cdp;
650 	char nodename[20];
651 	int i;
652 
653 	/*
654 	 * two ways here:
655 	 *
656 	 * the "old" way (debug.name -> value) which was emulated by
657 	 * the sysctl(8) binary
658 	 *
659 	 * the new way, which the sysctl(8) binary was actually using
660 
661 	 node	debug
662 	 node	debug.0
663 	 string debug.0.name
664 	 int	debug.0.value
665 	 int	debug.name
666 
667 	 */
668 
669 	for (i = 0; i < CTL_DEBUG_MAXID; i++) {
670 		cdp = debugvars[i];
671 		if (cdp->debugname == NULL || cdp->debugvar == NULL)
672 			continue;
673 
674 		snprintf(nodename, sizeof(nodename), "debug%d", i);
675 		sysctl_createv(clog, 0, NULL, NULL,
676 			       CTLFLAG_PERMANENT|CTLFLAG_HIDDEN,
677 			       CTLTYPE_NODE, nodename, NULL,
678 			       NULL, 0, NULL, 0,
679 			       CTL_DEBUG, i, CTL_EOL);
680 		sysctl_createv(clog, 0, NULL, NULL,
681 			       CTLFLAG_PERMANENT|CTLFLAG_HIDDEN,
682 			       CTLTYPE_STRING, "name", NULL,
683 			       /*XXXUNCONST*/
684 			       NULL, 0, __UNCONST(cdp->debugname), 0,
685 			       CTL_DEBUG, i, CTL_DEBUG_NAME, CTL_EOL);
686 		sysctl_createv(clog, 0, NULL, NULL,
687 			       CTLFLAG_PERMANENT|CTLFLAG_HIDDEN,
688 			       CTLTYPE_INT, "value", NULL,
689 			       NULL, 0, cdp->debugvar, 0,
690 			       CTL_DEBUG, i, CTL_DEBUG_VALUE, CTL_EOL);
691 		sysctl_createv(clog, 0, NULL, NULL,
692 			       CTLFLAG_PERMANENT,
693 			       CTLTYPE_INT, cdp->debugname, NULL,
694 			       NULL, 0, cdp->debugvar, 0,
695 			       CTL_DEBUG, CTL_CREATE, CTL_EOL);
696 	}
697 }
698 #endif /* DEBUG */
699 
700 /*
701  * ********************************************************************
702  * section 2: private node-specific helper routines.
703  * ********************************************************************
704  */
705 
706 #ifdef DIAGNOSTIC
707 static int
708 sysctl_kern_trigger_panic(SYSCTLFN_ARGS)
709 {
710 	int newtrig, error;
711 	struct sysctlnode node;
712 
713 	newtrig = 0;
714 	node = *rnode;
715 	node.sysctl_data = &newtrig;
716 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
717 	if (error || newp == NULL)
718 		return (error);
719 
720 	if (newtrig != 0)
721 		panic("Panic triggered");
722 
723 	return (error);
724 }
725 #endif
726 
727 /*
728  * sysctl helper routine for kern.maxvnodes.  Drain vnodes if
729  * new value is lower than desiredvnodes and then calls reinit
730  * routines that needs to adjust to the new value.
731  */
732 static int
733 sysctl_kern_maxvnodes(SYSCTLFN_ARGS)
734 {
735 	int error, new_vnodes, old_vnodes, new_max;
736 	struct sysctlnode node;
737 
738 	new_vnodes = desiredvnodes;
739 	node = *rnode;
740 	node.sysctl_data = &new_vnodes;
741 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
742 	if (error || newp == NULL)
743 		return (error);
744 
745 	/*
746 	 * sysctl passes down unsigned values, require them
747 	 * to be positive
748 	 */
749 	if (new_vnodes <= 0)
750 		return (EINVAL);
751 
752 	/* Limits: 75% of kmem and physical memory. */
753 	new_max = calc_cache_size(vmem_size(kmem_arena, VMEM_FREE|VMEM_ALLOC),
754 	    75, 75) / VNODE_COST;
755 	if (new_vnodes > new_max)
756 		new_vnodes = new_max;
757 
758 	old_vnodes = desiredvnodes;
759 	desiredvnodes = new_vnodes;
760 	error = vfs_drainvnodes();
761 	if (error) {
762 		desiredvnodes = old_vnodes;
763 		return (error);
764 	}
765 	vfs_reinit();
766 	nchreinit();
767 
768 	return (0);
769 }
770 
771 /*
772  * sysctl helper routine for kern.messages.
773  * Alters boothowto to display kernel messages in increasing verbosity
774  * from 0 to 4.
775  */
776 
777 #define MAXMESSAGES            4
778 static int
779 sysctl_kern_messages(SYSCTLFN_ARGS)
780 {
781 	int error, messageverbose, messagemask, newboothowto;
782 	struct sysctlnode node;
783 
784 	messagemask = (AB_NORMAL|AB_QUIET|AB_SILENT|AB_VERBOSE|AB_DEBUG);
785 	switch (boothowto & messagemask) {
786 	case AB_SILENT:
787 		messageverbose = 0;
788 		break;
789 	case AB_QUIET:
790 		messageverbose = 1;
791 		break;
792 	case AB_VERBOSE:
793 		messageverbose = 3;
794 		break;
795 	case AB_DEBUG:
796 		messageverbose = 4;
797 		break;
798 	case AB_NORMAL:
799 	default:
800 		messageverbose = 2;
801 }
802 
803 	node = *rnode;
804 	node.sysctl_data = &messageverbose;
805 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
806 	if (error || newp == NULL)
807 		return (error);
808 	if (messageverbose < 0 || messageverbose > MAXMESSAGES)
809 		return EINVAL;
810 
811 	/* Set boothowto */
812 	newboothowto = boothowto & ~messagemask;
813 
814 	switch (messageverbose) {
815 	case 0:
816 		newboothowto |= AB_SILENT;
817 		break;
818 	case 1:
819 		newboothowto |= AB_QUIET;
820 		break;
821 	case 3:
822 		newboothowto |= AB_VERBOSE;
823 		break;
824 	case 4:
825 		newboothowto |= AB_DEBUG;
826 		break;
827 	case 2:
828 	default:                /* Messages default to normal. */
829 		break;
830 	}
831 
832 	boothowto = newboothowto;
833 
834 	return (0);
835 }
836 
837 /*
838  * sysctl helper routine for rtc_offset - set time after changes
839  */
840 static int
841 sysctl_kern_rtc_offset(SYSCTLFN_ARGS)
842 {
843 	struct timespec ts, delta;
844 	int error, new_rtc_offset;
845 	struct sysctlnode node;
846 
847 	new_rtc_offset = rtc_offset;
848 	node = *rnode;
849 	node.sysctl_data = &new_rtc_offset;
850 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
851 	if (error || newp == NULL)
852 		return (error);
853 
854 	if (kauth_authorize_system(l->l_cred, KAUTH_SYSTEM_TIME,
855 	    KAUTH_REQ_SYSTEM_TIME_RTCOFFSET,
856 	    KAUTH_ARG(new_rtc_offset), NULL, NULL))
857 		return (EPERM);
858 	if (rtc_offset == new_rtc_offset)
859 		return (0);
860 
861 	/* if we change the offset, adjust the time */
862 	nanotime(&ts);
863 	delta.tv_sec = 60 * (new_rtc_offset - rtc_offset);
864 	delta.tv_nsec = 0;
865 	timespecadd(&ts, &delta, &ts);
866 	rtc_offset = new_rtc_offset;
867 	return (settime(l->l_proc, &ts));
868 }
869 
870 /*
871  * sysctl helper routine for kern.maxproc. Ensures that the new
872  * values are not too low or too high.
873  */
874 static int
875 sysctl_kern_maxproc(SYSCTLFN_ARGS)
876 {
877 	int error, nmaxproc;
878 	struct sysctlnode node;
879 
880 	nmaxproc = maxproc;
881 	node = *rnode;
882 	node.sysctl_data = &nmaxproc;
883 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
884 	if (error || newp == NULL)
885 		return (error);
886 
887 	if (nmaxproc < 0 || nmaxproc >= PID_MAX)
888 		return (EINVAL);
889 #ifdef __HAVE_CPU_MAXPROC
890 	if (nmaxproc > cpu_maxproc())
891 		return (EINVAL);
892 #endif
893 	maxproc = nmaxproc;
894 
895 	return (0);
896 }
897 
898 /*
899  * sysctl helper function for kern.hostid. The hostid is a long, but
900  * we export it as an int, so we need to give it a little help.
901  */
902 static int
903 sysctl_kern_hostid(SYSCTLFN_ARGS)
904 {
905 	int error, inthostid;
906 	struct sysctlnode node;
907 
908 	inthostid = hostid;  /* XXX assumes sizeof int <= sizeof long */
909 	node = *rnode;
910 	node.sysctl_data = &inthostid;
911 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
912 	if (error || newp == NULL)
913 		return (error);
914 
915 	hostid = (unsigned)inthostid;
916 
917 	return (0);
918 }
919 
920 /*
921  * sysctl helper routine for kern.defcorename. In the case of a new
922  * string being assigned, check that it's not a zero-length string.
923  * (XXX the check in -current doesn't work, but do we really care?)
924  */
925 static int
926 sysctl_kern_defcorename(SYSCTLFN_ARGS)
927 {
928 	int error;
929 	char *newcorename;
930 	struct sysctlnode node;
931 
932 	newcorename = PNBUF_GET();
933 	node = *rnode;
934 	node.sysctl_data = &newcorename[0];
935 	memcpy(node.sysctl_data, rnode->sysctl_data, MAXPATHLEN);
936 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
937 	if (error || newp == NULL) {
938 		goto done;
939 	}
940 
941 	/*
942 	 * when sysctl_lookup() deals with a string, it's guaranteed
943 	 * to come back nul terminated. So there.  :)
944 	 */
945 	if (strlen(newcorename) == 0) {
946 		error = EINVAL;
947 	} else {
948 		memcpy(rnode->sysctl_data, node.sysctl_data, MAXPATHLEN);
949 		error = 0;
950 	}
951 done:
952 	PNBUF_PUT(newcorename);
953 	return error;
954 }
955 
956 /*
957  * sysctl helper routine for kern.cp_time node. Adds up cpu time
958  * across all cpus.
959  */
960 static int
961 sysctl_kern_cptime(SYSCTLFN_ARGS)
962 {
963 	struct sysctlnode node = *rnode;
964 	uint64_t *cp_time = NULL;
965 	int error, n = ncpu, i;
966 	struct cpu_info *ci;
967 	CPU_INFO_ITERATOR cii;
968 
969 	/*
970 	 * if you specifically pass a buffer that is the size of the
971 	 * sum, or if you are probing for the size, you get the "sum"
972 	 * of cp_time (and the size thereof) across all processors.
973 	 *
974 	 * alternately, you can pass an additional mib number and get
975 	 * cp_time for that particular processor.
976 	 */
977 	switch (namelen) {
978 	case 0:
979 		if (*oldlenp == sizeof(uint64_t) * CPUSTATES || oldp == NULL) {
980 			node.sysctl_size = sizeof(uint64_t) * CPUSTATES;
981 			n = -1; /* SUM */
982 		}
983 		else {
984 			node.sysctl_size = n * sizeof(uint64_t) * CPUSTATES;
985 			n = -2; /* ALL */
986 		}
987 		break;
988 	case 1:
989 		if (name[0] < 0 || name[0] >= n)
990 			return (ENOENT); /* ENOSUCHPROCESSOR */
991 		node.sysctl_size = sizeof(uint64_t) * CPUSTATES;
992 		n = name[0];
993 		/*
994 		 * adjust these so that sysctl_lookup() will be happy
995 		 */
996 		name++;
997 		namelen--;
998 		break;
999 	default:
1000 		return (EINVAL);
1001 	}
1002 
1003 	cp_time = kmem_alloc(node.sysctl_size, KM_SLEEP);
1004 	node.sysctl_data = cp_time;
1005 	memset(cp_time, 0, node.sysctl_size);
1006 
1007 	for (CPU_INFO_FOREACH(cii, ci)) {
1008 		if (n <= 0) {
1009 			for (i = 0; i < CPUSTATES; i++) {
1010 				cp_time[i] += ci->ci_schedstate.spc_cp_time[i];
1011 			}
1012 		}
1013 		/*
1014 		 * if a specific processor was requested and we just
1015 		 * did it, we're done here
1016 		 */
1017 		if (n == 0)
1018 			break;
1019 		/*
1020 		 * if doing "all", skip to next cp_time set for next processor
1021 		 */
1022 		if (n == -2)
1023 			cp_time += CPUSTATES;
1024 		/*
1025 		 * if we're doing a specific processor, we're one
1026 		 * processor closer
1027 		 */
1028 		if (n > 0)
1029 			n--;
1030 	}
1031 
1032 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1033 	kmem_free(node.sysctl_data, node.sysctl_size);
1034 	return (error);
1035 }
1036 
1037 #if NPTY > 0
1038 /*
1039  * sysctl helper routine for kern.maxptys. Ensures that any new value
1040  * is acceptable to the pty subsystem.
1041  */
1042 static int
1043 sysctl_kern_maxptys(SYSCTLFN_ARGS)
1044 {
1045 	int pty_maxptys(int, int);		/* defined in kern/tty_pty.c */
1046 	int error, xmax;
1047 	struct sysctlnode node;
1048 
1049 	/* get current value of maxptys */
1050 	xmax = pty_maxptys(0, 0);
1051 
1052 	node = *rnode;
1053 	node.sysctl_data = &xmax;
1054 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1055 	if (error || newp == NULL)
1056 		return (error);
1057 
1058 	if (xmax != pty_maxptys(xmax, 1))
1059 		return (EINVAL);
1060 
1061 	return (0);
1062 }
1063 #endif /* NPTY > 0 */
1064 
1065 /*
1066  * sysctl helper routine to do kern.lwp.* work.
1067  */
1068 static int
1069 sysctl_kern_lwp(SYSCTLFN_ARGS)
1070 {
1071 	struct kinfo_lwp klwp;
1072 	struct proc *p;
1073 	struct lwp *l2, *l3;
1074 	char *where, *dp;
1075 	int pid, elem_size, elem_count;
1076 	int buflen, needed, error;
1077 	bool gotit;
1078 
1079 	if (namelen == 1 && name[0] == CTL_QUERY)
1080 		return (sysctl_query(SYSCTLFN_CALL(rnode)));
1081 
1082 	dp = where = oldp;
1083 	buflen = where != NULL ? *oldlenp : 0;
1084 	error = needed = 0;
1085 
1086 	if (newp != NULL || namelen != 3)
1087 		return (EINVAL);
1088 	pid = name[0];
1089 	elem_size = name[1];
1090 	elem_count = name[2];
1091 
1092 	sysctl_unlock();
1093 	if (pid == -1) {
1094 		mutex_enter(proc_lock);
1095 		PROCLIST_FOREACH(p, &allproc) {
1096 			/* Grab a hold on the process. */
1097 			if (!rw_tryenter(&p->p_reflock, RW_READER)) {
1098 				continue;
1099 			}
1100 			mutex_exit(proc_lock);
1101 
1102 			mutex_enter(p->p_lock);
1103 			LIST_FOREACH(l2, &p->p_lwps, l_sibling) {
1104 				if (buflen >= elem_size && elem_count > 0) {
1105 					lwp_lock(l2);
1106 					fill_lwp(l2, &klwp);
1107 					lwp_unlock(l2);
1108 					mutex_exit(p->p_lock);
1109 
1110 					/*
1111 					 * Copy out elem_size, but not
1112 					 * larger than the size of a
1113 					 * 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,
1123 					    l_sibling) {
1124 						if (l2 == l3)
1125 							break;
1126 					}
1127 					if (l3 == NULL) {
1128 						mutex_exit(p->p_lock);
1129 						rw_exit(&p->p_reflock);
1130 						error = EAGAIN;
1131 						goto cleanup;
1132 					}
1133 					dp += elem_size;
1134 					buflen -= elem_size;
1135 					elem_count--;
1136 				}
1137 				needed += elem_size;
1138 			}
1139 			mutex_exit(p->p_lock);
1140 
1141 			/* Drop reference to process. */
1142 			mutex_enter(proc_lock);
1143 			rw_exit(&p->p_reflock);
1144 		}
1145 		mutex_exit(proc_lock);
1146 	} else {
1147 		mutex_enter(proc_lock);
1148 		p = proc_find(pid);
1149 		if (p == NULL) {
1150 			error = ESRCH;
1151 			mutex_exit(proc_lock);
1152 			goto cleanup;
1153 		}
1154 		/* Grab a hold on the process. */
1155 		gotit = rw_tryenter(&p->p_reflock, RW_READER);
1156 		mutex_exit(proc_lock);
1157 		if (!gotit) {
1158 			error = ESRCH;
1159 			goto cleanup;
1160 		}
1161 
1162 		mutex_enter(p->p_lock);
1163 		LIST_FOREACH(l2, &p->p_lwps, l_sibling) {
1164 			if (buflen >= elem_size && elem_count > 0) {
1165 				lwp_lock(l2);
1166 				fill_lwp(l2, &klwp);
1167 				lwp_unlock(l2);
1168 				mutex_exit(p->p_lock);
1169 				/*
1170 				 * Copy out elem_size, but not larger than
1171 				 * the size of a struct kinfo_proc2.
1172 				 */
1173 				error = dcopyout(l, &klwp, dp,
1174 				    min(sizeof(klwp), elem_size));
1175 				if (error) {
1176 					rw_exit(&p->p_reflock);
1177 					goto cleanup;
1178 				}
1179 				mutex_enter(p->p_lock);
1180 				LIST_FOREACH(l3, &p->p_lwps, l_sibling) {
1181 					if (l2 == l3)
1182 						break;
1183 				}
1184 				if (l3 == NULL) {
1185 					mutex_exit(p->p_lock);
1186 					rw_exit(&p->p_reflock);
1187 					error = EAGAIN;
1188 					goto cleanup;
1189 				}
1190 				dp += elem_size;
1191 				buflen -= elem_size;
1192 				elem_count--;
1193 			}
1194 			needed += elem_size;
1195 		}
1196 		mutex_exit(p->p_lock);
1197 
1198 		/* Drop reference to process. */
1199 		rw_exit(&p->p_reflock);
1200 	}
1201 
1202 	if (where != NULL) {
1203 		*oldlenp = dp - where;
1204 		if (needed > *oldlenp) {
1205 			sysctl_relock();
1206 			return (ENOMEM);
1207 		}
1208 	} else {
1209 		needed += KERN_LWPSLOP;
1210 		*oldlenp = needed;
1211 	}
1212 	error = 0;
1213  cleanup:
1214 	sysctl_relock();
1215 	return (error);
1216 }
1217 
1218 /*
1219  * sysctl helper routine for kern.forkfsleep node. Ensures that the
1220  * given value is not too large or two small, and is at least one
1221  * timer tick if not zero.
1222  */
1223 static int
1224 sysctl_kern_forkfsleep(SYSCTLFN_ARGS)
1225 {
1226 	/* userland sees value in ms, internally is in ticks */
1227 	extern int forkfsleep;		/* defined in kern/kern_fork.c */
1228 	int error, timo, lsleep;
1229 	struct sysctlnode node;
1230 
1231 	lsleep = forkfsleep * 1000 / hz;
1232 	node = *rnode;
1233 	node.sysctl_data = &lsleep;
1234 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1235 	if (error || newp == NULL)
1236 		return (error);
1237 
1238 	/* refuse negative values, and overly 'long time' */
1239 	if (lsleep < 0 || lsleep > MAXSLP * 1000)
1240 		return (EINVAL);
1241 
1242 	timo = mstohz(lsleep);
1243 
1244 	/* if the interval is >0 ms && <1 tick, use 1 tick */
1245 	if (lsleep != 0 && timo == 0)
1246 		forkfsleep = 1;
1247 	else
1248 		forkfsleep = timo;
1249 
1250 	return (0);
1251 }
1252 
1253 /*
1254  * sysctl helper routine for kern.root_partition
1255  */
1256 static int
1257 sysctl_kern_root_partition(SYSCTLFN_ARGS)
1258 {
1259 	int rootpart = DISKPART(rootdev);
1260 	struct sysctlnode node = *rnode;
1261 
1262 	node.sysctl_data = &rootpart;
1263 	return (sysctl_lookup(SYSCTLFN_CALL(&node)));
1264 }
1265 
1266 /*
1267  * sysctl helper function for kern.drivers
1268  */
1269 static int
1270 sysctl_kern_drivers(SYSCTLFN_ARGS)
1271 {
1272 	int error;
1273 	size_t buflen;
1274 	struct kinfo_drivers kd;
1275 	char *start, *where;
1276 	const char *dname;
1277 	int i;
1278 	extern struct devsw_conv *devsw_conv;
1279 	extern int max_devsw_convs;
1280 
1281 	start = where = oldp;
1282 	buflen = *oldlenp;
1283 	if (where == NULL) {
1284 		*oldlenp = max_devsw_convs * sizeof kd;
1285 		return 0;
1286 	}
1287 
1288 	/*
1289 	 * An array of kinfo_drivers structures
1290 	 */
1291 	error = 0;
1292 	sysctl_unlock();
1293 	mutex_enter(&device_lock);
1294 	for (i = 0; i < max_devsw_convs; i++) {
1295 		dname = devsw_conv[i].d_name;
1296 		if (dname == NULL)
1297 			continue;
1298 		if (buflen < sizeof kd) {
1299 			error = ENOMEM;
1300 			break;
1301 		}
1302 		memset(&kd, 0, sizeof(kd));
1303 		kd.d_bmajor = devsw_conv[i].d_bmajor;
1304 		kd.d_cmajor = devsw_conv[i].d_cmajor;
1305 		strlcpy(kd.d_name, dname, sizeof kd.d_name);
1306 		mutex_exit(&device_lock);
1307 		error = dcopyout(l, &kd, where, sizeof kd);
1308 		mutex_enter(&device_lock);
1309 		if (error != 0)
1310 			break;
1311 		buflen -= sizeof kd;
1312 		where += sizeof kd;
1313 	}
1314 	mutex_exit(&device_lock);
1315 	sysctl_relock();
1316 	*oldlenp = where - start;
1317 	return error;
1318 }
1319 
1320 static int
1321 sysctl_security_setidcore(SYSCTLFN_ARGS)
1322 {
1323 	int newsize, error;
1324 	struct sysctlnode node;
1325 
1326 	node = *rnode;
1327 	node.sysctl_data = &newsize;
1328 	newsize = *(int *)rnode->sysctl_data;
1329 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1330 	if (error || newp == NULL)
1331 		return error;
1332 
1333 	if (kauth_authorize_system(l->l_cred, KAUTH_SYSTEM_SETIDCORE,
1334 	    0, NULL, NULL, NULL))
1335 		return (EPERM);
1336 
1337 	*(int *)rnode->sysctl_data = newsize;
1338 
1339 	return 0;
1340 }
1341 
1342 static int
1343 sysctl_security_setidcorename(SYSCTLFN_ARGS)
1344 {
1345 	int error;
1346 	char *newsetidcorename;
1347 	struct sysctlnode node;
1348 
1349 	newsetidcorename = PNBUF_GET();
1350 	node = *rnode;
1351 	node.sysctl_data = newsetidcorename;
1352 	memcpy(node.sysctl_data, rnode->sysctl_data, MAXPATHLEN);
1353 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1354 	if (error || newp == NULL) {
1355 		goto out;
1356 	}
1357 	if (kauth_authorize_system(l->l_cred, KAUTH_SYSTEM_SETIDCORE,
1358 	    0, NULL, NULL, NULL)) {
1359 		error = EPERM;
1360 		goto out;
1361 	}
1362 	if (strlen(newsetidcorename) == 0) {
1363 		error = EINVAL;
1364 		goto out;
1365 	}
1366 	memcpy(rnode->sysctl_data, node.sysctl_data, MAXPATHLEN);
1367 out:
1368 	PNBUF_PUT(newsetidcorename);
1369 	return error;
1370 }
1371 
1372 /*
1373  * sysctl helper routine for kern.cp_id node. Maps cpus to their
1374  * cpuids.
1375  */
1376 static int
1377 sysctl_kern_cpid(SYSCTLFN_ARGS)
1378 {
1379 	struct sysctlnode node = *rnode;
1380 	uint64_t *cp_id = NULL;
1381 	int error, n = ncpu;
1382 	struct cpu_info *ci;
1383 	CPU_INFO_ITERATOR cii;
1384 
1385 	/*
1386 	 * Here you may either retrieve a single cpu id or the whole
1387 	 * set. The size you get back when probing depends on what
1388 	 * you ask for.
1389 	 */
1390 	switch (namelen) {
1391 	case 0:
1392 		node.sysctl_size = n * sizeof(uint64_t);
1393 		n = -2; /* ALL */
1394 		break;
1395 	case 1:
1396 		if (name[0] < 0 || name[0] >= n)
1397 			return (ENOENT); /* ENOSUCHPROCESSOR */
1398 		node.sysctl_size = sizeof(uint64_t);
1399 		n = name[0];
1400 		/*
1401 		 * adjust these so that sysctl_lookup() will be happy
1402 		 */
1403 		name++;
1404 		namelen--;
1405 		break;
1406 	default:
1407 		return (EINVAL);
1408 	}
1409 
1410 	cp_id = kmem_alloc(node.sysctl_size, KM_SLEEP);
1411 	node.sysctl_data = cp_id;
1412 	memset(cp_id, 0, node.sysctl_size);
1413 
1414 	for (CPU_INFO_FOREACH(cii, ci)) {
1415 		if (n <= 0)
1416 			cp_id[0] = cpu_index(ci);
1417 		/*
1418 		 * if a specific processor was requested and we just
1419 		 * did it, we're done here
1420 		 */
1421 		if (n == 0)
1422 			break;
1423 		/*
1424 		 * if doing "all", skip to next cp_id slot for next processor
1425 		 */
1426 		if (n == -2)
1427 			cp_id++;
1428 		/*
1429 		 * if we're doing a specific processor, we're one
1430 		 * processor closer
1431 		 */
1432 		if (n > 0)
1433 			n--;
1434 	}
1435 
1436 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1437 	kmem_free(node.sysctl_data, node.sysctl_size);
1438 	return (error);
1439 }
1440 
1441 /*
1442  * sysctl helper routine for hw.usermem and hw.usermem64. Values are
1443  * calculate on the fly taking into account integer overflow and the
1444  * current wired count.
1445  */
1446 static int
1447 sysctl_hw_usermem(SYSCTLFN_ARGS)
1448 {
1449 	u_int ui;
1450 	u_quad_t uq;
1451 	struct sysctlnode node;
1452 
1453 	node = *rnode;
1454 	switch (rnode->sysctl_num) {
1455 	case HW_USERMEM:
1456 		if ((ui = physmem - uvmexp.wired) > (UINT_MAX / PAGE_SIZE))
1457 			ui = UINT_MAX;
1458 		else
1459 			ui *= PAGE_SIZE;
1460 		node.sysctl_data = &ui;
1461 		break;
1462 	case HW_USERMEM64:
1463 		uq = (u_quad_t)(physmem - uvmexp.wired) * PAGE_SIZE;
1464 		node.sysctl_data = &uq;
1465 		break;
1466 	default:
1467 		return (EINVAL);
1468 	}
1469 
1470 	return (sysctl_lookup(SYSCTLFN_CALL(&node)));
1471 }
1472 
1473 /*
1474  * sysctl helper routine for kern.cnmagic node. Pulls the old value
1475  * out, encoded, and stuffs the new value in for decoding.
1476  */
1477 static int
1478 sysctl_hw_cnmagic(SYSCTLFN_ARGS)
1479 {
1480 	char magic[CNS_LEN];
1481 	int error;
1482 	struct sysctlnode node;
1483 
1484 	if (oldp)
1485 		cn_get_magic(magic, CNS_LEN);
1486 	node = *rnode;
1487 	node.sysctl_data = &magic[0];
1488 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1489 	if (error || newp == NULL)
1490 		return (error);
1491 
1492 	return (cn_set_magic(magic));
1493 }
1494 
1495 /*
1496  * ********************************************************************
1497  * section 3: public helper routines that are used for more than one
1498  * node
1499  * ********************************************************************
1500  */
1501 
1502 /*
1503  * sysctl helper routine for the kern.root_device node and some ports'
1504  * machdep.root_device nodes.
1505  */
1506 int
1507 sysctl_root_device(SYSCTLFN_ARGS)
1508 {
1509 	struct sysctlnode node;
1510 
1511 	node = *rnode;
1512 	node.sysctl_data = __UNCONST(device_xname(root_device));
1513 	node.sysctl_size = strlen(device_xname(root_device)) + 1;
1514 	return (sysctl_lookup(SYSCTLFN_CALL(&node)));
1515 }
1516 
1517 /*
1518  * sysctl helper routine for kern.consdev, dependent on the current
1519  * state of the console. Also used for machdep.console_device on some
1520  * ports.
1521  */
1522 int
1523 sysctl_consdev(SYSCTLFN_ARGS)
1524 {
1525 	dev_t consdev;
1526 	uint32_t oconsdev;
1527 	struct sysctlnode node;
1528 
1529 	if (cn_tab != NULL)
1530 		consdev = cn_tab->cn_dev;
1531 	else
1532 		consdev = NODEV;
1533 	node = *rnode;
1534 	switch (*oldlenp) {
1535 	case sizeof(consdev):
1536 		node.sysctl_data = &consdev;
1537 		node.sysctl_size = sizeof(consdev);
1538 		break;
1539 	case sizeof(oconsdev):
1540 		oconsdev = (uint32_t)consdev;
1541 		node.sysctl_data = &oconsdev;
1542 		node.sysctl_size = sizeof(oconsdev);
1543 		break;
1544 	default:
1545 		return EINVAL;
1546 	}
1547 	return (sysctl_lookup(SYSCTLFN_CALL(&node)));
1548 }
1549 
1550 /*
1551  * ********************************************************************
1552  * section 4: support for some helpers
1553  * ********************************************************************
1554  */
1555 
1556 
1557 /*
1558  * Fill in a kinfo_lwp structure for the specified lwp.
1559  */
1560 static void
1561 fill_lwp(struct lwp *l, struct kinfo_lwp *kl)
1562 {
1563 	struct proc *p = l->l_proc;
1564 	struct timeval tv;
1565 
1566 	KASSERT(lwp_locked(l, NULL));
1567 
1568 	memset(kl, 0, sizeof(*kl));
1569 
1570 	kl->l_forw = 0;
1571 	kl->l_back = 0;
1572 	kl->l_laddr = PTRTOUINT64(l);
1573 	kl->l_addr = PTRTOUINT64(l->l_addr);
1574 	kl->l_stat = l->l_stat;
1575 	kl->l_lid = l->l_lid;
1576 	kl->l_flag = L_INMEM;
1577 	kl->l_flag |= sysctl_map_flags(sysctl_lwpprflagmap, l->l_prflag);
1578 	kl->l_flag |= sysctl_map_flags(sysctl_lwpflagmap, l->l_flag);
1579 
1580 	kl->l_swtime = l->l_swtime;
1581 	kl->l_slptime = l->l_slptime;
1582 	if (l->l_stat == LSONPROC)
1583 		kl->l_schedflags = l->l_cpu->ci_schedstate.spc_flags;
1584 	else
1585 		kl->l_schedflags = 0;
1586 	kl->l_priority = lwp_eprio(l);
1587 	kl->l_usrpri = l->l_priority;
1588 	if (l->l_wchan)
1589 		strncpy(kl->l_wmesg, l->l_wmesg, sizeof(kl->l_wmesg));
1590 	kl->l_wchan = PTRTOUINT64(l->l_wchan);
1591 	kl->l_cpuid = cpu_index(l->l_cpu);
1592 	bintime2timeval(&l->l_rtime, &tv);
1593 	kl->l_rtime_sec = tv.tv_sec;
1594 	kl->l_rtime_usec = tv.tv_usec;
1595 	kl->l_cpticks = l->l_cpticks;
1596 	kl->l_pctcpu = l->l_pctcpu;
1597 	kl->l_pid = p->p_pid;
1598 	if (l->l_name == NULL)
1599 		kl->l_name[0] = '\0';
1600 	else
1601 		strlcpy(kl->l_name, l->l_name, sizeof(kl->l_name));
1602 }
1603