xref: /netbsd-src/sys/kern/init_sysctl.c (revision 404fbe5fb94ca1e054339640cabb2801ce52dd30)
1 /*	$NetBSD: init_sysctl.c,v 1.153 2009/01/11 02:45:51 christos Exp $ */
2 
3 /*-
4  * Copyright (c) 2003, 2007, 2008 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 1.153 2009/01/11 02:45:51 christos Exp $");
34 
35 #include "opt_sysv.h"
36 #include "opt_compat_netbsd32.h"
37 #include "opt_compat_netbsd.h"
38 #include "opt_sa.h"
39 #include "opt_posix.h"
40 #include "pty.h"
41 #include "rnd.h"
42 
43 #include <sys/types.h>
44 #include <sys/param.h>
45 #include <sys/sysctl.h>
46 #include <sys/cpu.h>
47 #include <sys/errno.h>
48 #include <sys/systm.h>
49 #include <sys/kernel.h>
50 #include <sys/unistd.h>
51 #include <sys/disklabel.h>
52 #include <sys/rnd.h>
53 #include <sys/vnode.h>
54 #include <sys/mount.h>
55 #include <sys/namei.h>
56 #include <sys/msgbuf.h>
57 #include <dev/cons.h>
58 #include <sys/socketvar.h>
59 #include <sys/file.h>
60 #include <sys/filedesc.h>
61 #include <sys/tty.h>
62 #include <sys/malloc.h>
63 #include <sys/resource.h>
64 #include <sys/resourcevar.h>
65 #include <sys/exec.h>
66 #include <sys/conf.h>
67 #include <sys/device.h>
68 #include <sys/stat.h>
69 #include <sys/kauth.h>
70 #include <sys/ktrace.h>
71 #include <sys/ksem.h>
72 
73 #ifdef COMPAT_NETBSD32
74 #include <compat/netbsd32/netbsd32.h>
75 #endif
76 #ifdef COMPAT_50
77 #include <compat/sys/time.h>
78 #endif
79 
80 #ifdef KERN_SA
81 #include <sys/sa.h>
82 #endif
83 
84 #include <sys/cpu.h>
85 
86 #if defined(MODULAR) || defined(P1003_1B_SEMAPHORE)
87 int posix_semaphores = 200112;
88 #else
89 int posix_semaphores;
90 #endif
91 
92 int security_setidcore_dump;
93 char security_setidcore_path[MAXPATHLEN] = "/var/crash/%n.core";
94 uid_t security_setidcore_owner = 0;
95 gid_t security_setidcore_group = 0;
96 mode_t security_setidcore_mode = (S_IRUSR|S_IWUSR);
97 
98 static const u_int sysctl_flagmap[] = {
99 	PK_ADVLOCK, P_ADVLOCK,
100 	PK_EXEC, P_EXEC,
101 	PK_NOCLDWAIT, P_NOCLDWAIT,
102 	PK_32, P_32,
103 	PK_CLDSIGIGN, P_CLDSIGIGN,
104 	PK_SUGID, P_SUGID,
105 	0
106 };
107 
108 static const u_int sysctl_sflagmap[] = {
109 	PS_NOCLDSTOP, P_NOCLDSTOP,
110 	PS_WEXIT, P_WEXIT,
111 	PS_STOPFORK, P_STOPFORK,
112 	PS_STOPEXEC, P_STOPEXEC,
113 	PS_STOPEXIT, P_STOPEXIT,
114 	0
115 };
116 
117 static const u_int sysctl_slflagmap[] = {
118 	PSL_TRACED, P_TRACED,
119 	PSL_FSTRACE, P_FSTRACE,
120 	PSL_CHTRACED, P_CHTRACED,
121 	PSL_SYSCALL, P_SYSCALL,
122 	0
123 };
124 
125 static const u_int sysctl_lflagmap[] = {
126 	PL_CONTROLT, P_CONTROLT,
127 	PL_PPWAIT, P_PPWAIT,
128 	0
129 };
130 
131 static const u_int sysctl_stflagmap[] = {
132 	PST_PROFIL, P_PROFIL,
133 	0
134 
135 };
136 
137 static const u_int sysctl_lwpflagmap[] = {
138 	LW_INMEM, P_INMEM,
139 	LW_SINTR, P_SINTR,
140 	LW_SYSTEM, P_SYSTEM,
141 	LW_SA, P_SA,	/* WRS ??? */
142 	0
143 };
144 
145 static const u_int sysctl_lwpprflagmap[] = {
146 	LPR_DETACHED, L_DETACHED,
147 	0
148 };
149 
150 /*
151  * try over estimating by 5 procs/lwps
152  */
153 #define KERN_PROCSLOP	(5 * sizeof(struct kinfo_proc))
154 #define KERN_LWPSLOP	(5 * sizeof(struct kinfo_lwp))
155 
156 static int dcopyout(struct lwp *, const void *, void *, size_t);
157 
158 static int
159 dcopyout(struct lwp *l, const void *kaddr, void *uaddr, size_t len)
160 {
161 	int error;
162 
163 	error = copyout(kaddr, uaddr, len);
164 	ktrmibio(-1, UIO_READ, uaddr, len, error);
165 
166 	return error;
167 }
168 
169 #ifdef DIAGNOSTIC
170 static int sysctl_kern_trigger_panic(SYSCTLFN_PROTO);
171 #endif
172 static int sysctl_kern_maxvnodes(SYSCTLFN_PROTO);
173 static int sysctl_kern_rtc_offset(SYSCTLFN_PROTO);
174 static int sysctl_kern_maxproc(SYSCTLFN_PROTO);
175 static int sysctl_kern_hostid(SYSCTLFN_PROTO);
176 static int sysctl_setlen(SYSCTLFN_PROTO);
177 static int sysctl_kern_clockrate(SYSCTLFN_PROTO);
178 static int sysctl_kern_file(SYSCTLFN_PROTO);
179 static int sysctl_msgbuf(SYSCTLFN_PROTO);
180 static int sysctl_kern_defcorename(SYSCTLFN_PROTO);
181 static int sysctl_kern_cptime(SYSCTLFN_PROTO);
182 #if NPTY > 0
183 static int sysctl_kern_maxptys(SYSCTLFN_PROTO);
184 #endif /* NPTY > 0 */
185 static int sysctl_kern_sbmax(SYSCTLFN_PROTO);
186 static int sysctl_kern_urnd(SYSCTLFN_PROTO);
187 static int sysctl_kern_arnd(SYSCTLFN_PROTO);
188 static int sysctl_kern_lwp(SYSCTLFN_PROTO);
189 static int sysctl_kern_forkfsleep(SYSCTLFN_PROTO);
190 static int sysctl_kern_root_partition(SYSCTLFN_PROTO);
191 static int sysctl_kern_drivers(SYSCTLFN_PROTO);
192 static int sysctl_kern_file2(SYSCTLFN_PROTO);
193 static int sysctl_security_setidcore(SYSCTLFN_PROTO);
194 static int sysctl_security_setidcorename(SYSCTLFN_PROTO);
195 static int sysctl_kern_cpid(SYSCTLFN_PROTO);
196 static int sysctl_doeproc(SYSCTLFN_PROTO);
197 static int sysctl_kern_proc_args(SYSCTLFN_PROTO);
198 static int sysctl_hw_usermem(SYSCTLFN_PROTO);
199 static int sysctl_hw_cnmagic(SYSCTLFN_PROTO);
200 
201 static u_int sysctl_map_flags(const u_int *, u_int);
202 static void fill_kproc2(struct proc *, struct kinfo_proc2 *, bool);
203 static void fill_lwp(struct lwp *l, struct kinfo_lwp *kl);
204 static void fill_file(struct kinfo_file *, const file_t *, const fdfile_t *,
205 		      int, pid_t);
206 
207 /*
208  * ********************************************************************
209  * section 1: setup routines
210  * ********************************************************************
211  * These functions are stuffed into a link set for sysctl setup
212  * functions. They're never called or referenced from anywhere else.
213  * ********************************************************************
214  */
215 
216 /*
217  * sets up the base nodes...
218  */
219 SYSCTL_SETUP(sysctl_root_setup, "sysctl base setup")
220 {
221 
222 	sysctl_createv(clog, 0, NULL, NULL,
223 		       CTLFLAG_PERMANENT,
224 		       CTLTYPE_NODE, "kern",
225 		       SYSCTL_DESCR("High kernel"),
226 		       NULL, 0, NULL, 0,
227 		       CTL_KERN, CTL_EOL);
228 	sysctl_createv(clog, 0, NULL, NULL,
229 		       CTLFLAG_PERMANENT,
230 		       CTLTYPE_NODE, "vm",
231 		       SYSCTL_DESCR("Virtual memory"),
232 		       NULL, 0, NULL, 0,
233 		       CTL_VM, CTL_EOL);
234 	sysctl_createv(clog, 0, NULL, NULL,
235 		       CTLFLAG_PERMANENT,
236 		       CTLTYPE_NODE, "vfs",
237 		       SYSCTL_DESCR("Filesystem"),
238 		       NULL, 0, NULL, 0,
239 		       CTL_VFS, CTL_EOL);
240 	sysctl_createv(clog, 0, NULL, NULL,
241 		       CTLFLAG_PERMANENT,
242 		       CTLTYPE_NODE, "net",
243 		       SYSCTL_DESCR("Networking"),
244 		       NULL, 0, NULL, 0,
245 		       CTL_NET, CTL_EOL);
246 	sysctl_createv(clog, 0, NULL, NULL,
247 		       CTLFLAG_PERMANENT,
248 		       CTLTYPE_NODE, "debug",
249 		       SYSCTL_DESCR("Debugging"),
250 		       NULL, 0, NULL, 0,
251 		       CTL_DEBUG, CTL_EOL);
252 	sysctl_createv(clog, 0, NULL, NULL,
253 		       CTLFLAG_PERMANENT,
254 		       CTLTYPE_NODE, "hw",
255 		       SYSCTL_DESCR("Generic CPU, I/O"),
256 		       NULL, 0, NULL, 0,
257 		       CTL_HW, CTL_EOL);
258 	sysctl_createv(clog, 0, NULL, NULL,
259 		       CTLFLAG_PERMANENT,
260 		       CTLTYPE_NODE, "machdep",
261 		       SYSCTL_DESCR("Machine dependent"),
262 		       NULL, 0, NULL, 0,
263 		       CTL_MACHDEP, CTL_EOL);
264 	/*
265 	 * this node is inserted so that the sysctl nodes in libc can
266 	 * operate.
267 	 */
268 	sysctl_createv(clog, 0, NULL, NULL,
269 		       CTLFLAG_PERMANENT,
270 		       CTLTYPE_NODE, "user",
271 		       SYSCTL_DESCR("User-level"),
272 		       NULL, 0, NULL, 0,
273 		       CTL_USER, CTL_EOL);
274 	sysctl_createv(clog, 0, NULL, NULL,
275 		       CTLFLAG_PERMANENT,
276 		       CTLTYPE_NODE, "ddb",
277 		       SYSCTL_DESCR("In-kernel debugger"),
278 		       NULL, 0, NULL, 0,
279 		       CTL_DDB, CTL_EOL);
280 	sysctl_createv(clog, 0, NULL, NULL,
281 		       CTLFLAG_PERMANENT,
282 		       CTLTYPE_NODE, "proc",
283 		       SYSCTL_DESCR("Per-process"),
284 		       NULL, 0, NULL, 0,
285 		       CTL_PROC, CTL_EOL);
286 	sysctl_createv(clog, 0, NULL, NULL,
287 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
288 		       CTLTYPE_NODE, "vendor",
289 		       SYSCTL_DESCR("Vendor specific"),
290 		       NULL, 0, NULL, 0,
291 		       CTL_VENDOR, CTL_EOL);
292 	sysctl_createv(clog, 0, NULL, NULL,
293 		       CTLFLAG_PERMANENT,
294 		       CTLTYPE_NODE, "emul",
295 		       SYSCTL_DESCR("Emulation settings"),
296 		       NULL, 0, NULL, 0,
297 		       CTL_EMUL, CTL_EOL);
298 	sysctl_createv(clog, 0, NULL, NULL,
299 		       CTLFLAG_PERMANENT,
300 		       CTLTYPE_NODE, "security",
301 		       SYSCTL_DESCR("Security"),
302 		       NULL, 0, NULL, 0,
303 		       CTL_SECURITY, CTL_EOL);
304 }
305 
306 /*
307  * this setup routine is a replacement for kern_sysctl()
308  */
309 SYSCTL_SETUP(sysctl_kern_setup, "sysctl kern subtree setup")
310 {
311 	extern int kern_logsigexit;	/* defined in kern/kern_sig.c */
312 	extern fixpt_t ccpu;		/* defined in kern/kern_synch.c */
313 	extern int dumponpanic;		/* defined in kern/subr_prf.c */
314 	const struct sysctlnode *rnode;
315 
316 	sysctl_createv(clog, 0, NULL, NULL,
317 		       CTLFLAG_PERMANENT,
318 		       CTLTYPE_NODE, "kern", NULL,
319 		       NULL, 0, NULL, 0,
320 		       CTL_KERN, CTL_EOL);
321 
322 	sysctl_createv(clog, 0, NULL, NULL,
323 		       CTLFLAG_PERMANENT,
324 		       CTLTYPE_STRING, "ostype",
325 		       SYSCTL_DESCR("Operating system type"),
326 		       NULL, 0, &ostype, 0,
327 		       CTL_KERN, KERN_OSTYPE, CTL_EOL);
328 	sysctl_createv(clog, 0, NULL, NULL,
329 		       CTLFLAG_PERMANENT,
330 		       CTLTYPE_STRING, "osrelease",
331 		       SYSCTL_DESCR("Operating system release"),
332 		       NULL, 0, &osrelease, 0,
333 		       CTL_KERN, KERN_OSRELEASE, CTL_EOL);
334 	sysctl_createv(clog, 0, NULL, NULL,
335 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
336 		       CTLTYPE_INT, "osrevision",
337 		       SYSCTL_DESCR("Operating system revision"),
338 		       NULL, __NetBSD_Version__, NULL, 0,
339 		       CTL_KERN, KERN_OSREV, CTL_EOL);
340 	sysctl_createv(clog, 0, NULL, NULL,
341 		       CTLFLAG_PERMANENT,
342 		       CTLTYPE_STRING, "version",
343 		       SYSCTL_DESCR("Kernel version"),
344 		       NULL, 0, &version, 0,
345 		       CTL_KERN, KERN_VERSION, CTL_EOL);
346 	sysctl_createv(clog, 0, NULL, NULL,
347 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
348 		       CTLTYPE_INT, "maxvnodes",
349 		       SYSCTL_DESCR("Maximum number of vnodes"),
350 		       sysctl_kern_maxvnodes, 0, NULL, 0,
351 		       CTL_KERN, KERN_MAXVNODES, CTL_EOL);
352 	sysctl_createv(clog, 0, NULL, NULL,
353 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
354 		       CTLTYPE_INT, "maxproc",
355 		       SYSCTL_DESCR("Maximum number of simultaneous processes"),
356 		       sysctl_kern_maxproc, 0, NULL, 0,
357 		       CTL_KERN, KERN_MAXPROC, CTL_EOL);
358 	sysctl_createv(clog, 0, NULL, NULL,
359 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
360 		       CTLTYPE_INT, "maxfiles",
361 		       SYSCTL_DESCR("Maximum number of open files"),
362 		       NULL, 0, &maxfiles, 0,
363 		       CTL_KERN, KERN_MAXFILES, CTL_EOL);
364 	sysctl_createv(clog, 0, NULL, NULL,
365 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
366 		       CTLTYPE_INT, "argmax",
367 		       SYSCTL_DESCR("Maximum number of bytes of arguments to "
368 				    "execve(2)"),
369 		       NULL, ARG_MAX, NULL, 0,
370 		       CTL_KERN, KERN_ARGMAX, CTL_EOL);
371 	sysctl_createv(clog, 0, NULL, NULL,
372 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
373 		       CTLTYPE_STRING, "hostname",
374 		       SYSCTL_DESCR("System hostname"),
375 		       sysctl_setlen, 0, &hostname, MAXHOSTNAMELEN,
376 		       CTL_KERN, KERN_HOSTNAME, CTL_EOL);
377 	sysctl_createv(clog, 0, NULL, NULL,
378 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE|CTLFLAG_HEX,
379 		       CTLTYPE_INT, "hostid",
380 		       SYSCTL_DESCR("System host ID number"),
381 		       sysctl_kern_hostid, 0, NULL, 0,
382 		       CTL_KERN, KERN_HOSTID, CTL_EOL);
383 	sysctl_createv(clog, 0, NULL, NULL,
384 		       CTLFLAG_PERMANENT,
385 		       CTLTYPE_STRUCT, "clockrate",
386 		       SYSCTL_DESCR("Kernel clock rates"),
387 		       sysctl_kern_clockrate, 0, NULL,
388 		       sizeof(struct clockinfo),
389 		       CTL_KERN, KERN_CLOCKRATE, CTL_EOL);
390 	sysctl_createv(clog, 0, NULL, NULL,
391 		       CTLFLAG_PERMANENT,
392 		       CTLTYPE_INT, "hardclock_ticks",
393 		       SYSCTL_DESCR("Number of hardclock ticks"),
394 		       NULL, 0, &hardclock_ticks, sizeof(hardclock_ticks),
395 		       CTL_KERN, KERN_HARDCLOCK_TICKS, CTL_EOL);
396 	sysctl_createv(clog, 0, NULL, NULL,
397 		       CTLFLAG_PERMANENT,
398 		       CTLTYPE_STRUCT, "vnode",
399 		       SYSCTL_DESCR("System vnode table"),
400 		       sysctl_kern_vnode, 0, NULL, 0,
401 		       CTL_KERN, KERN_VNODE, CTL_EOL);
402 	sysctl_createv(clog, 0, NULL, NULL,
403 		       CTLFLAG_PERMANENT,
404 		       CTLTYPE_STRUCT, "file",
405 		       SYSCTL_DESCR("System open file table"),
406 		       sysctl_kern_file, 0, NULL, 0,
407 		       CTL_KERN, KERN_FILE, CTL_EOL);
408 #ifndef GPROF
409 	sysctl_createv(clog, 0, NULL, NULL,
410 		       CTLFLAG_PERMANENT,
411 		       CTLTYPE_NODE, "profiling",
412 		       SYSCTL_DESCR("Profiling information (not available)"),
413 		       sysctl_notavail, 0, NULL, 0,
414 		       CTL_KERN, KERN_PROF, CTL_EOL);
415 #endif
416 	sysctl_createv(clog, 0, NULL, NULL,
417 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
418 		       CTLTYPE_INT, "posix1version",
419 		       SYSCTL_DESCR("Version of ISO/IEC 9945 (POSIX 1003.1) "
420 				    "with which the operating system attempts "
421 				    "to comply"),
422 		       NULL, _POSIX_VERSION, NULL, 0,
423 		       CTL_KERN, KERN_POSIX1, CTL_EOL);
424 	sysctl_createv(clog, 0, NULL, NULL,
425 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
426 		       CTLTYPE_INT, "ngroups",
427 		       SYSCTL_DESCR("Maximum number of supplemental groups"),
428 		       NULL, NGROUPS_MAX, NULL, 0,
429 		       CTL_KERN, KERN_NGROUPS, CTL_EOL);
430 	sysctl_createv(clog, 0, NULL, NULL,
431 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
432 		       CTLTYPE_INT, "job_control",
433 		       SYSCTL_DESCR("Whether job control is available"),
434 		       NULL, 1, NULL, 0,
435 		       CTL_KERN, KERN_JOB_CONTROL, CTL_EOL);
436 	sysctl_createv(clog, 0, NULL, NULL,
437 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
438 		       CTLTYPE_INT, "saved_ids",
439 		       SYSCTL_DESCR("Whether POSIX saved set-group/user ID is "
440 				    "available"), NULL,
441 #ifdef _POSIX_SAVED_IDS
442 		       1,
443 #else /* _POSIX_SAVED_IDS */
444 		       0,
445 #endif /* _POSIX_SAVED_IDS */
446 		       NULL, 0, CTL_KERN, KERN_SAVED_IDS, CTL_EOL);
447 	sysctl_createv(clog, 0, NULL, NULL,
448 		       CTLFLAG_PERMANENT,
449 		       CTLTYPE_STRUCT, "boottime",
450 		       SYSCTL_DESCR("System boot time"),
451 		       NULL, 0, &boottime, sizeof(boottime),
452 		       CTL_KERN, KERN_BOOTTIME, CTL_EOL);
453 #ifdef COMPAT_50
454 	{
455 		extern struct timeval50 boottime50;
456 		sysctl_createv(clog, 0, NULL, NULL,
457 			       CTLFLAG_PERMANENT,
458 			       CTLTYPE_STRUCT, "oboottime",
459 			       SYSCTL_DESCR("System boot time"),
460 			       NULL, 0, &boottime50, sizeof(boottime50),
461 			       CTL_KERN, KERN_OBOOTTIME, CTL_EOL);
462 	}
463 #endif
464 	sysctl_createv(clog, 0, NULL, NULL,
465 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
466 		       CTLTYPE_STRING, "domainname",
467 		       SYSCTL_DESCR("YP domain name"),
468 		       sysctl_setlen, 0, &domainname, MAXHOSTNAMELEN,
469 		       CTL_KERN, KERN_DOMAINNAME, CTL_EOL);
470 	sysctl_createv(clog, 0, NULL, NULL,
471 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
472 		       CTLTYPE_INT, "maxpartitions",
473 		       SYSCTL_DESCR("Maximum number of partitions allowed per "
474 				    "disk"),
475 		       NULL, MAXPARTITIONS, NULL, 0,
476 		       CTL_KERN, KERN_MAXPARTITIONS, CTL_EOL);
477 	sysctl_createv(clog, 0, NULL, NULL,
478 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
479 		       CTLTYPE_INT, "rawpartition",
480 		       SYSCTL_DESCR("Raw partition of a disk"),
481 		       NULL, RAW_PART, NULL, 0,
482 		       CTL_KERN, KERN_RAWPARTITION, CTL_EOL);
483 	sysctl_createv(clog, 0, NULL, NULL,
484 		       CTLFLAG_PERMANENT,
485 		       CTLTYPE_STRUCT, "timex", NULL,
486 		       sysctl_notavail, 0, NULL, 0,
487 		       CTL_KERN, KERN_TIMEX, CTL_EOL);
488 	sysctl_createv(clog, 0, NULL, NULL,
489 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
490 		       CTLTYPE_INT, "rtc_offset",
491 		       SYSCTL_DESCR("Offset of real time clock from UTC in "
492 				    "minutes"),
493 		       sysctl_kern_rtc_offset, 0, &rtc_offset, 0,
494 		       CTL_KERN, KERN_RTC_OFFSET, CTL_EOL);
495 	sysctl_createv(clog, 0, NULL, NULL,
496 		       CTLFLAG_PERMANENT,
497 		       CTLTYPE_STRING, "root_device",
498 		       SYSCTL_DESCR("Name of the root device"),
499 		       sysctl_root_device, 0, NULL, 0,
500 		       CTL_KERN, KERN_ROOT_DEVICE, CTL_EOL);
501 	sysctl_createv(clog, 0, NULL, NULL,
502 		       CTLFLAG_PERMANENT,
503 		       CTLTYPE_INT, "msgbufsize",
504 		       SYSCTL_DESCR("Size of the kernel message buffer"),
505 		       sysctl_msgbuf, 0, NULL, 0,
506 		       CTL_KERN, KERN_MSGBUFSIZE, CTL_EOL);
507 	sysctl_createv(clog, 0, NULL, NULL,
508 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
509 		       CTLTYPE_INT, "fsync",
510 		       SYSCTL_DESCR("Whether the POSIX 1003.1b File "
511 				    "Synchronization Option is available on "
512 				    "this system"),
513 		       NULL, 1, NULL, 0,
514 		       CTL_KERN, KERN_FSYNC, CTL_EOL);
515 	sysctl_createv(clog, 0, NULL, NULL,
516 		       CTLFLAG_PERMANENT,
517 		       CTLTYPE_NODE, "ipc",
518 		       SYSCTL_DESCR("SysV IPC options"),
519 		       NULL, 0, NULL, 0,
520 		       CTL_KERN, KERN_SYSVIPC, CTL_EOL);
521 	sysctl_createv(clog, 0, NULL, NULL,
522 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
523 		       CTLTYPE_INT, "sysvmsg",
524 		       SYSCTL_DESCR("System V style message support available"),
525 		       NULL,
526 #ifdef SYSVMSG
527 		       1,
528 #else /* SYSVMSG */
529 		       0,
530 #endif /* SYSVMSG */
531 		       NULL, 0, CTL_KERN, KERN_SYSVIPC, KERN_SYSVIPC_MSG, CTL_EOL);
532 	sysctl_createv(clog, 0, NULL, NULL,
533 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
534 		       CTLTYPE_INT, "sysvsem",
535 		       SYSCTL_DESCR("System V style semaphore support "
536 				    "available"), NULL,
537 #ifdef SYSVSEM
538 		       1,
539 #else /* SYSVSEM */
540 		       0,
541 #endif /* SYSVSEM */
542 		       NULL, 0, CTL_KERN, KERN_SYSVIPC, KERN_SYSVIPC_SEM, CTL_EOL);
543 	sysctl_createv(clog, 0, NULL, NULL,
544 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
545 		       CTLTYPE_INT, "sysvshm",
546 		       SYSCTL_DESCR("System V style shared memory support "
547 				    "available"), NULL,
548 #ifdef SYSVSHM
549 		       1,
550 #else /* SYSVSHM */
551 		       0,
552 #endif /* SYSVSHM */
553 		       NULL, 0, CTL_KERN, KERN_SYSVIPC, KERN_SYSVIPC_SHM, CTL_EOL);
554 	sysctl_createv(clog, 0, NULL, NULL,
555 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
556 		       CTLTYPE_INT, "synchronized_io",
557 		       SYSCTL_DESCR("Whether the POSIX 1003.1b Synchronized "
558 				    "I/O Option is available on this system"),
559 		       NULL, 1, NULL, 0,
560 		       CTL_KERN, KERN_SYNCHRONIZED_IO, CTL_EOL);
561 	sysctl_createv(clog, 0, NULL, NULL,
562 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
563 		       CTLTYPE_INT, "iov_max",
564 		       SYSCTL_DESCR("Maximum number of iovec structures per "
565 				    "process"),
566 		       NULL, IOV_MAX, NULL, 0,
567 		       CTL_KERN, KERN_IOV_MAX, CTL_EOL);
568 	sysctl_createv(clog, 0, NULL, NULL,
569 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
570 		       CTLTYPE_INT, "mapped_files",
571 		       SYSCTL_DESCR("Whether the POSIX 1003.1b Memory Mapped "
572 				    "Files Option is available on this system"),
573 		       NULL, 1, NULL, 0,
574 		       CTL_KERN, KERN_MAPPED_FILES, CTL_EOL);
575 	sysctl_createv(clog, 0, NULL, NULL,
576 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
577 		       CTLTYPE_INT, "memlock",
578 		       SYSCTL_DESCR("Whether the POSIX 1003.1b Process Memory "
579 				    "Locking Option is available on this "
580 				    "system"),
581 		       NULL, 1, NULL, 0,
582 		       CTL_KERN, KERN_MEMLOCK, CTL_EOL);
583 	sysctl_createv(clog, 0, NULL, NULL,
584 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
585 		       CTLTYPE_INT, "memlock_range",
586 		       SYSCTL_DESCR("Whether the POSIX 1003.1b Range Memory "
587 				    "Locking Option is available on this "
588 				    "system"),
589 		       NULL, 1, NULL, 0,
590 		       CTL_KERN, KERN_MEMLOCK_RANGE, CTL_EOL);
591 	sysctl_createv(clog, 0, NULL, NULL,
592 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
593 		       CTLTYPE_INT, "memory_protection",
594 		       SYSCTL_DESCR("Whether the POSIX 1003.1b Memory "
595 				    "Protection Option is available on this "
596 				    "system"),
597 		       NULL, 1, NULL, 0,
598 		       CTL_KERN, KERN_MEMORY_PROTECTION, CTL_EOL);
599 	sysctl_createv(clog, 0, NULL, NULL,
600 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
601 		       CTLTYPE_INT, "login_name_max",
602 		       SYSCTL_DESCR("Maximum login name length"),
603 		       NULL, LOGIN_NAME_MAX, NULL, 0,
604 		       CTL_KERN, KERN_LOGIN_NAME_MAX, CTL_EOL);
605 	sysctl_createv(clog, 0, NULL, NULL,
606 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
607 		       CTLTYPE_STRING, "defcorename",
608 		       SYSCTL_DESCR("Default core file name"),
609 		       sysctl_kern_defcorename, 0, defcorename, MAXPATHLEN,
610 		       CTL_KERN, KERN_DEFCORENAME, CTL_EOL);
611 	sysctl_createv(clog, 0, NULL, NULL,
612 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
613 		       CTLTYPE_INT, "logsigexit",
614 		       SYSCTL_DESCR("Log process exit when caused by signals"),
615 		       NULL, 0, &kern_logsigexit, 0,
616 		       CTL_KERN, KERN_LOGSIGEXIT, CTL_EOL);
617 	sysctl_createv(clog, 0, NULL, NULL,
618 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
619 		       CTLTYPE_INT, "fscale",
620 		       SYSCTL_DESCR("Kernel fixed-point scale factor"),
621 		       NULL, FSCALE, NULL, 0,
622 		       CTL_KERN, KERN_FSCALE, CTL_EOL);
623 	sysctl_createv(clog, 0, NULL, NULL,
624 		       CTLFLAG_PERMANENT,
625 		       CTLTYPE_INT, "ccpu",
626 		       SYSCTL_DESCR("Scheduler exponential decay value"),
627 		       NULL, 0, &ccpu, 0,
628 		       CTL_KERN, KERN_CCPU, CTL_EOL);
629 	sysctl_createv(clog, 0, NULL, NULL,
630 		       CTLFLAG_PERMANENT,
631 		       CTLTYPE_STRUCT, "cp_time",
632 		       SYSCTL_DESCR("Clock ticks spent in different CPU states"),
633 		       sysctl_kern_cptime, 0, NULL, 0,
634 		       CTL_KERN, KERN_CP_TIME, CTL_EOL);
635 	sysctl_createv(clog, 0, NULL, NULL,
636 		       CTLFLAG_PERMANENT,
637 		       CTLTYPE_INT, "msgbuf",
638 		       SYSCTL_DESCR("Kernel message buffer"),
639 		       sysctl_msgbuf, 0, NULL, 0,
640 		       CTL_KERN, KERN_MSGBUF, CTL_EOL);
641 	sysctl_createv(clog, 0, NULL, NULL,
642 		       CTLFLAG_PERMANENT,
643 		       CTLTYPE_STRUCT, "consdev",
644 		       SYSCTL_DESCR("Console device"),
645 		       sysctl_consdev, 0, NULL, sizeof(dev_t),
646 		       CTL_KERN, KERN_CONSDEV, CTL_EOL);
647 #if NPTY > 0
648 	sysctl_createv(clog, 0, NULL, NULL,
649 		       CTLFLAG_PERMANENT,
650 		       CTLTYPE_INT, "maxptys",
651 		       SYSCTL_DESCR("Maximum number of pseudo-ttys"),
652 		       sysctl_kern_maxptys, 0, NULL, 0,
653 		       CTL_KERN, KERN_MAXPTYS, CTL_EOL);
654 #endif /* NPTY > 0 */
655 	sysctl_createv(clog, 0, NULL, NULL,
656 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
657 		       CTLTYPE_INT, "maxphys",
658 		       SYSCTL_DESCR("Maximum raw I/O transfer size"),
659 		       NULL, MAXPHYS, NULL, 0,
660 		       CTL_KERN, KERN_MAXPHYS, CTL_EOL);
661 	sysctl_createv(clog, 0, NULL, NULL,
662 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
663 		       CTLTYPE_INT, "sbmax",
664 		       SYSCTL_DESCR("Maximum socket buffer size"),
665 		       sysctl_kern_sbmax, 0, NULL, 0,
666 		       CTL_KERN, KERN_SBMAX, CTL_EOL);
667 	sysctl_createv(clog, 0, NULL, NULL,
668 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
669 		       CTLTYPE_INT, "monotonic_clock",
670 		       SYSCTL_DESCR("Implementation version of the POSIX "
671 				    "1003.1b Monotonic Clock Option"),
672 		       /* XXX _POSIX_VERSION */
673 		       NULL, _POSIX_MONOTONIC_CLOCK, NULL, 0,
674 		       CTL_KERN, KERN_MONOTONIC_CLOCK, CTL_EOL);
675 	sysctl_createv(clog, 0, NULL, NULL,
676 		       CTLFLAG_PERMANENT,
677 		       CTLTYPE_INT, "urandom",
678 		       SYSCTL_DESCR("Random integer value"),
679 		       sysctl_kern_urnd, 0, NULL, 0,
680 		       CTL_KERN, KERN_URND, CTL_EOL);
681 	sysctl_createv(clog, 0, NULL, NULL,
682 		       CTLFLAG_PERMANENT,
683 		       CTLTYPE_INT, "arandom",
684 		       SYSCTL_DESCR("n bytes of random data"),
685 		       sysctl_kern_arnd, 0, NULL, 0,
686 		       CTL_KERN, KERN_ARND, CTL_EOL);
687 	sysctl_createv(clog, 0, NULL, NULL,
688 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
689 		       CTLTYPE_INT, "labelsector",
690 		       SYSCTL_DESCR("Sector number containing the disklabel"),
691 		       NULL, LABELSECTOR, NULL, 0,
692 		       CTL_KERN, KERN_LABELSECTOR, CTL_EOL);
693 	sysctl_createv(clog, 0, NULL, NULL,
694 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
695 		       CTLTYPE_INT, "labeloffset",
696 		       SYSCTL_DESCR("Offset of the disklabel within the "
697 				    "sector"),
698 		       NULL, LABELOFFSET, NULL, 0,
699 		       CTL_KERN, KERN_LABELOFFSET, CTL_EOL);
700 	sysctl_createv(clog, 0, NULL, NULL,
701 		       CTLFLAG_PERMANENT,
702 		       CTLTYPE_NODE, "lwp",
703 		       SYSCTL_DESCR("System-wide LWP information"),
704 		       sysctl_kern_lwp, 0, NULL, 0,
705 		       CTL_KERN, KERN_LWP, CTL_EOL);
706 	sysctl_createv(clog, 0, NULL, NULL,
707 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
708 		       CTLTYPE_INT, "forkfsleep",
709 		       SYSCTL_DESCR("Milliseconds to sleep on fork failure due "
710 				    "to process limits"),
711 		       sysctl_kern_forkfsleep, 0, NULL, 0,
712 		       CTL_KERN, KERN_FORKFSLEEP, CTL_EOL);
713 	sysctl_createv(clog, 0, NULL, NULL,
714 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
715 		       CTLTYPE_INT, "posix_threads",
716 		       SYSCTL_DESCR("Version of IEEE Std 1003.1 and its "
717 				    "Threads option to which the system "
718 				    "attempts to conform"),
719 		       /* XXX _POSIX_VERSION */
720 		       NULL, _POSIX_THREADS, NULL, 0,
721 		       CTL_KERN, KERN_POSIX_THREADS, CTL_EOL);
722 	sysctl_createv(clog, 0, NULL, NULL,
723 		       CTLFLAG_PERMANENT,
724 		       CTLTYPE_INT, "posix_semaphores",
725 		       SYSCTL_DESCR("Version of IEEE Std 1003.1 and its "
726 				    "Semaphores option to which the system "
727 				    "attempts to conform"), NULL,
728 		       0, &posix_semaphores,
729 		       0, CTL_KERN, KERN_POSIX_SEMAPHORES, CTL_EOL);
730 	sysctl_createv(clog, 0, NULL, NULL,
731 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
732 		       CTLTYPE_INT, "posix_barriers",
733 		       SYSCTL_DESCR("Version of IEEE Std 1003.1 and its "
734 				    "Barriers option to which the system "
735 				    "attempts to conform"),
736 		       /* XXX _POSIX_VERSION */
737 		       NULL, _POSIX_BARRIERS, NULL, 0,
738 		       CTL_KERN, KERN_POSIX_BARRIERS, CTL_EOL);
739 	sysctl_createv(clog, 0, NULL, NULL,
740 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
741 		       CTLTYPE_INT, "posix_timers",
742 		       SYSCTL_DESCR("Version of IEEE Std 1003.1 and its "
743 				    "Timers option to which the system "
744 				    "attempts to conform"),
745 		       /* XXX _POSIX_VERSION */
746 		       NULL, _POSIX_TIMERS, NULL, 0,
747 		       CTL_KERN, KERN_POSIX_TIMERS, CTL_EOL);
748 	sysctl_createv(clog, 0, NULL, NULL,
749 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
750 		       CTLTYPE_INT, "posix_spin_locks",
751 		       SYSCTL_DESCR("Version of IEEE Std 1003.1 and its Spin "
752 				    "Locks option to which the system attempts "
753 				    "to conform"),
754 		       /* XXX _POSIX_VERSION */
755 		       NULL, _POSIX_SPIN_LOCKS, NULL, 0,
756 		       CTL_KERN, KERN_POSIX_SPIN_LOCKS, CTL_EOL);
757 	sysctl_createv(clog, 0, NULL, NULL,
758 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
759 		       CTLTYPE_INT, "posix_reader_writer_locks",
760 		       SYSCTL_DESCR("Version of IEEE Std 1003.1 and its "
761 				    "Read-Write Locks option to which the "
762 				    "system attempts to conform"),
763 		       /* XXX _POSIX_VERSION */
764 		       NULL, _POSIX_READER_WRITER_LOCKS, NULL, 0,
765 		       CTL_KERN, KERN_POSIX_READER_WRITER_LOCKS, CTL_EOL);
766 	sysctl_createv(clog, 0, NULL, NULL,
767 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
768 		       CTLTYPE_INT, "dump_on_panic",
769 		       SYSCTL_DESCR("Perform a crash dump on system panic"),
770 		       NULL, 0, &dumponpanic, 0,
771 		       CTL_KERN, KERN_DUMP_ON_PANIC, CTL_EOL);
772 #ifdef DIAGNOSTIC
773 	sysctl_createv(clog, 0, NULL, NULL,
774 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
775 		       CTLTYPE_INT, "panic_now",
776 		       SYSCTL_DESCR("Trigger a panic"),
777 		       sysctl_kern_trigger_panic, 0, NULL, 0,
778 		       CTL_KERN, CTL_CREATE, CTL_EOL);
779 #endif
780 	sysctl_createv(clog, 0, NULL, NULL,
781 		       CTLFLAG_PERMANENT,
782 		       CTLTYPE_INT, "root_partition",
783 		       SYSCTL_DESCR("Root partition on the root device"),
784 		       sysctl_kern_root_partition, 0, NULL, 0,
785 		       CTL_KERN, KERN_ROOT_PARTITION, CTL_EOL);
786 	sysctl_createv(clog, 0, NULL, NULL,
787 		       CTLFLAG_PERMANENT,
788 		       CTLTYPE_STRUCT, "drivers",
789 		       SYSCTL_DESCR("List of all drivers with block and "
790 				    "character device numbers"),
791 		       sysctl_kern_drivers, 0, NULL, 0,
792 		       CTL_KERN, KERN_DRIVERS, CTL_EOL);
793 	sysctl_createv(clog, 0, NULL, NULL,
794 		       CTLFLAG_PERMANENT,
795 		       CTLTYPE_STRUCT, "file2",
796 		       SYSCTL_DESCR("System open file table"),
797 		       sysctl_kern_file2, 0, NULL, 0,
798 		       CTL_KERN, KERN_FILE2, CTL_EOL);
799 	sysctl_createv(clog, 0, NULL, NULL,
800 		       CTLFLAG_PERMANENT,
801 		       CTLTYPE_STRUCT, "cp_id",
802 		       SYSCTL_DESCR("Mapping of CPU number to CPU id"),
803 		       sysctl_kern_cpid, 0, NULL, 0,
804 		       CTL_KERN, KERN_CP_ID, CTL_EOL);
805 	sysctl_createv(clog, 0, NULL, &rnode,
806 		       CTLFLAG_PERMANENT,
807 		       CTLTYPE_NODE, "coredump",
808 		       SYSCTL_DESCR("Coredump settings."),
809 		       NULL, 0, NULL, 0,
810 		       CTL_KERN, CTL_CREATE, CTL_EOL);
811 	sysctl_createv(clog, 0, &rnode, &rnode,
812 		       CTLFLAG_PERMANENT,
813 		       CTLTYPE_NODE, "setid",
814 		       SYSCTL_DESCR("Set-id processes' coredump settings."),
815 		       NULL, 0, NULL, 0,
816 		       CTL_CREATE, CTL_EOL);
817 	sysctl_createv(clog, 0, &rnode, NULL,
818 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
819 		       CTLTYPE_INT, "dump",
820 		       SYSCTL_DESCR("Allow set-id processes to dump core."),
821 		       sysctl_security_setidcore, 0, &security_setidcore_dump,
822 		       sizeof(security_setidcore_dump),
823 		       CTL_CREATE, CTL_EOL);
824 	sysctl_createv(clog, 0, &rnode, NULL,
825 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
826 		       CTLTYPE_STRING, "path",
827 		       SYSCTL_DESCR("Path pattern for set-id coredumps."),
828 		       sysctl_security_setidcorename, 0,
829 		       &security_setidcore_path,
830 		       sizeof(security_setidcore_path),
831 		       CTL_CREATE, CTL_EOL);
832 	sysctl_createv(clog, 0, &rnode, NULL,
833 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
834 		       CTLTYPE_INT, "owner",
835 		       SYSCTL_DESCR("Owner id for set-id processes' cores."),
836 		       sysctl_security_setidcore, 0, &security_setidcore_owner,
837 		       0,
838 		       CTL_CREATE, CTL_EOL);
839 	sysctl_createv(clog, 0, &rnode, NULL,
840 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
841 		       CTLTYPE_INT, "group",
842 		       SYSCTL_DESCR("Group id for set-id processes' cores."),
843 		       sysctl_security_setidcore, 0, &security_setidcore_group,
844 		       0,
845 		       CTL_CREATE, CTL_EOL);
846 	sysctl_createv(clog, 0, &rnode, NULL,
847 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
848 		       CTLTYPE_INT, "mode",
849 		       SYSCTL_DESCR("Mode for set-id processes' cores."),
850 		       sysctl_security_setidcore, 0, &security_setidcore_mode,
851 		       0,
852 		       CTL_CREATE, CTL_EOL);
853 #ifdef KERN_SA
854 	sysctl_createv(clog, 0, NULL, NULL,
855 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
856 		       CTLTYPE_INT, "no_sa_support",
857 		       SYSCTL_DESCR("0 if the kernel supports SA, otherwise it doesn't"),
858 		       NULL, 0, &sa_system_disabled, 0,
859 		       CTL_KERN, CTL_CREATE, CTL_EOL);
860 #else
861 	sysctl_createv(clog, 0, NULL, NULL,
862 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
863 		       CTLTYPE_INT, "no_sa_support",
864 		       SYSCTL_DESCR("0 if the kernel supports SA, otherwise it doesn't"),
865 		       NULL, 1, NULL, 0,
866 		       CTL_KERN, CTL_CREATE, CTL_EOL);
867 #endif
868 
869 	/* kern.posix. */
870 	sysctl_createv(clog, 0, NULL, &rnode,
871 			CTLFLAG_PERMANENT,
872 			CTLTYPE_NODE, "posix",
873 			SYSCTL_DESCR("POSIX options"),
874 			NULL, 0, NULL, 0,
875 			CTL_KERN, CTL_CREATE, CTL_EOL);
876 	sysctl_createv(clog, 0, &rnode, NULL,
877 			CTLFLAG_PERMANENT | CTLFLAG_READWRITE,
878 			CTLTYPE_INT, "semmax",
879 			SYSCTL_DESCR("Maximal number of semaphores"),
880 			NULL, 0, &ksem_max, 0,
881 			CTL_CREATE, CTL_EOL);
882 }
883 
884 SYSCTL_SETUP(sysctl_kern_proc_setup,
885 	     "sysctl kern.proc/proc2/proc_args subtree setup")
886 {
887 
888 	sysctl_createv(clog, 0, NULL, NULL,
889 		       CTLFLAG_PERMANENT,
890 		       CTLTYPE_NODE, "kern", NULL,
891 		       NULL, 0, NULL, 0,
892 		       CTL_KERN, CTL_EOL);
893 
894 	sysctl_createv(clog, 0, NULL, NULL,
895 		       CTLFLAG_PERMANENT,
896 		       CTLTYPE_NODE, "proc",
897 		       SYSCTL_DESCR("System-wide process information"),
898 		       sysctl_doeproc, 0, NULL, 0,
899 		       CTL_KERN, KERN_PROC, CTL_EOL);
900 	sysctl_createv(clog, 0, NULL, NULL,
901 		       CTLFLAG_PERMANENT,
902 		       CTLTYPE_NODE, "proc2",
903 		       SYSCTL_DESCR("Machine-independent process information"),
904 		       sysctl_doeproc, 0, NULL, 0,
905 		       CTL_KERN, KERN_PROC2, CTL_EOL);
906 	sysctl_createv(clog, 0, NULL, NULL,
907 		       CTLFLAG_PERMANENT,
908 		       CTLTYPE_NODE, "proc_args",
909 		       SYSCTL_DESCR("Process argument information"),
910 		       sysctl_kern_proc_args, 0, NULL, 0,
911 		       CTL_KERN, KERN_PROC_ARGS, CTL_EOL);
912 
913 	/*
914 	  "nodes" under these:
915 
916 	  KERN_PROC_ALL
917 	  KERN_PROC_PID pid
918 	  KERN_PROC_PGRP pgrp
919 	  KERN_PROC_SESSION sess
920 	  KERN_PROC_TTY tty
921 	  KERN_PROC_UID uid
922 	  KERN_PROC_RUID uid
923 	  KERN_PROC_GID gid
924 	  KERN_PROC_RGID gid
925 
926 	  all in all, probably not worth the effort...
927 	*/
928 }
929 
930 SYSCTL_SETUP(sysctl_hw_setup, "sysctl hw subtree setup")
931 {
932 	u_int u;
933 	u_quad_t q;
934 
935 	sysctl_createv(clog, 0, NULL, NULL,
936 		       CTLFLAG_PERMANENT,
937 		       CTLTYPE_NODE, "hw", NULL,
938 		       NULL, 0, NULL, 0,
939 		       CTL_HW, CTL_EOL);
940 
941 	sysctl_createv(clog, 0, NULL, NULL,
942 		       CTLFLAG_PERMANENT,
943 		       CTLTYPE_STRING, "machine",
944 		       SYSCTL_DESCR("Machine class"),
945 		       NULL, 0, machine, 0,
946 		       CTL_HW, HW_MACHINE, CTL_EOL);
947 	sysctl_createv(clog, 0, NULL, NULL,
948 		       CTLFLAG_PERMANENT,
949 		       CTLTYPE_STRING, "model",
950 		       SYSCTL_DESCR("Machine model"),
951 		       NULL, 0, cpu_model, 0,
952 		       CTL_HW, HW_MODEL, CTL_EOL);
953 	sysctl_createv(clog, 0, NULL, NULL,
954 		       CTLFLAG_PERMANENT,
955 		       CTLTYPE_INT, "ncpu",
956 		       SYSCTL_DESCR("Number of CPUs configured"),
957 		       NULL, 0, &ncpu, 0,
958 		       CTL_HW, HW_NCPU, CTL_EOL);
959 	sysctl_createv(clog, 0, NULL, NULL,
960 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
961 		       CTLTYPE_INT, "byteorder",
962 		       SYSCTL_DESCR("System byte order"),
963 		       NULL, BYTE_ORDER, NULL, 0,
964 		       CTL_HW, HW_BYTEORDER, CTL_EOL);
965 	u = ((u_int)physmem > (UINT_MAX / PAGE_SIZE)) ?
966 		UINT_MAX : physmem * PAGE_SIZE;
967 	sysctl_createv(clog, 0, NULL, NULL,
968 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
969 		       CTLTYPE_INT, "physmem",
970 		       SYSCTL_DESCR("Bytes of physical memory"),
971 		       NULL, u, NULL, 0,
972 		       CTL_HW, HW_PHYSMEM, CTL_EOL);
973 	sysctl_createv(clog, 0, NULL, NULL,
974 		       CTLFLAG_PERMANENT,
975 		       CTLTYPE_INT, "usermem",
976 		       SYSCTL_DESCR("Bytes of non-kernel memory"),
977 		       sysctl_hw_usermem, 0, NULL, 0,
978 		       CTL_HW, HW_USERMEM, CTL_EOL);
979 	sysctl_createv(clog, 0, NULL, NULL,
980 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
981 		       CTLTYPE_INT, "pagesize",
982 		       SYSCTL_DESCR("Software page size"),
983 		       NULL, PAGE_SIZE, NULL, 0,
984 		       CTL_HW, HW_PAGESIZE, CTL_EOL);
985 	sysctl_createv(clog, 0, NULL, NULL,
986 		       CTLFLAG_PERMANENT,
987 		       CTLTYPE_STRING, "machine_arch",
988 		       SYSCTL_DESCR("Machine CPU class"),
989 		       NULL, 0, machine_arch, 0,
990 		       CTL_HW, HW_MACHINE_ARCH, CTL_EOL);
991 	sysctl_createv(clog, 0, NULL, NULL,
992 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
993 		       CTLTYPE_INT, "alignbytes",
994 		       SYSCTL_DESCR("Alignment constraint for all possible "
995 				    "data types"),
996 		       NULL, ALIGNBYTES, NULL, 0,
997 		       CTL_HW, HW_ALIGNBYTES, CTL_EOL);
998 	sysctl_createv(clog, 0, NULL, NULL,
999 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE|CTLFLAG_HEX,
1000 		       CTLTYPE_STRING, "cnmagic",
1001 		       SYSCTL_DESCR("Console magic key sequence"),
1002 		       sysctl_hw_cnmagic, 0, NULL, CNS_LEN,
1003 		       CTL_HW, HW_CNMAGIC, CTL_EOL);
1004 	q = (u_quad_t)physmem * PAGE_SIZE;
1005 	sysctl_createv(clog, 0, NULL, NULL,
1006 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
1007 		       CTLTYPE_QUAD, "physmem64",
1008 		       SYSCTL_DESCR("Bytes of physical memory"),
1009 		       NULL, q, NULL, 0,
1010 		       CTL_HW, HW_PHYSMEM64, CTL_EOL);
1011 	sysctl_createv(clog, 0, NULL, NULL,
1012 		       CTLFLAG_PERMANENT,
1013 		       CTLTYPE_QUAD, "usermem64",
1014 		       SYSCTL_DESCR("Bytes of non-kernel memory"),
1015 		       sysctl_hw_usermem, 0, NULL, 0,
1016 		       CTL_HW, HW_USERMEM64, CTL_EOL);
1017 	sysctl_createv(clog, 0, NULL, NULL,
1018 		       CTLFLAG_PERMANENT,
1019 		       CTLTYPE_INT, "ncpuonline",
1020 		       SYSCTL_DESCR("Number of CPUs online"),
1021 		       NULL, 0, &ncpuonline, 0,
1022 		       CTL_HW, HW_NCPUONLINE, CTL_EOL);
1023 }
1024 
1025 #ifdef DEBUG
1026 /*
1027  * Debugging related system variables.
1028  */
1029 struct ctldebug /* debug0, */ /* debug1, */ debug2, debug3, debug4;
1030 struct ctldebug debug5, debug6, debug7, debug8, debug9;
1031 struct ctldebug debug10, debug11, debug12, debug13, debug14;
1032 struct ctldebug debug15, debug16, debug17, debug18, debug19;
1033 static struct ctldebug *debugvars[CTL_DEBUG_MAXID] = {
1034 	&debug0, &debug1, &debug2, &debug3, &debug4,
1035 	&debug5, &debug6, &debug7, &debug8, &debug9,
1036 	&debug10, &debug11, &debug12, &debug13, &debug14,
1037 	&debug15, &debug16, &debug17, &debug18, &debug19,
1038 };
1039 
1040 /*
1041  * this setup routine is a replacement for debug_sysctl()
1042  *
1043  * note that it creates several nodes per defined debug variable
1044  */
1045 SYSCTL_SETUP(sysctl_debug_setup, "sysctl debug subtree setup")
1046 {
1047 	struct ctldebug *cdp;
1048 	char nodename[20];
1049 	int i;
1050 
1051 	/*
1052 	 * two ways here:
1053 	 *
1054 	 * the "old" way (debug.name -> value) which was emulated by
1055 	 * the sysctl(8) binary
1056 	 *
1057 	 * the new way, which the sysctl(8) binary was actually using
1058 
1059 	 node	debug
1060 	 node	debug.0
1061 	 string debug.0.name
1062 	 int	debug.0.value
1063 	 int	debug.name
1064 
1065 	 */
1066 
1067 	sysctl_createv(clog, 0, NULL, NULL,
1068 		       CTLFLAG_PERMANENT,
1069 		       CTLTYPE_NODE, "debug", NULL,
1070 		       NULL, 0, NULL, 0,
1071 		       CTL_DEBUG, CTL_EOL);
1072 
1073 	for (i = 0; i < CTL_DEBUG_MAXID; i++) {
1074 		cdp = debugvars[i];
1075 		if (cdp->debugname == NULL || cdp->debugvar == NULL)
1076 			continue;
1077 
1078 		snprintf(nodename, sizeof(nodename), "debug%d", i);
1079 		sysctl_createv(clog, 0, NULL, NULL,
1080 			       CTLFLAG_PERMANENT|CTLFLAG_HIDDEN,
1081 			       CTLTYPE_NODE, nodename, NULL,
1082 			       NULL, 0, NULL, 0,
1083 			       CTL_DEBUG, i, CTL_EOL);
1084 		sysctl_createv(clog, 0, NULL, NULL,
1085 			       CTLFLAG_PERMANENT|CTLFLAG_HIDDEN,
1086 			       CTLTYPE_STRING, "name", NULL,
1087 			       /*XXXUNCONST*/
1088 			       NULL, 0, __UNCONST(cdp->debugname), 0,
1089 			       CTL_DEBUG, i, CTL_DEBUG_NAME, CTL_EOL);
1090 		sysctl_createv(clog, 0, NULL, NULL,
1091 			       CTLFLAG_PERMANENT|CTLFLAG_HIDDEN,
1092 			       CTLTYPE_INT, "value", NULL,
1093 			       NULL, 0, cdp->debugvar, 0,
1094 			       CTL_DEBUG, i, CTL_DEBUG_VALUE, CTL_EOL);
1095 		sysctl_createv(clog, 0, NULL, NULL,
1096 			       CTLFLAG_PERMANENT,
1097 			       CTLTYPE_INT, cdp->debugname, NULL,
1098 			       NULL, 0, cdp->debugvar, 0,
1099 			       CTL_DEBUG, CTL_CREATE, CTL_EOL);
1100 	}
1101 }
1102 #endif /* DEBUG */
1103 
1104 /*
1105  * ********************************************************************
1106  * section 2: private node-specific helper routines.
1107  * ********************************************************************
1108  */
1109 
1110 #ifdef DIAGNOSTIC
1111 static int
1112 sysctl_kern_trigger_panic(SYSCTLFN_ARGS)
1113 {
1114 	int newtrig, error;
1115 	struct sysctlnode node;
1116 
1117 	newtrig = 0;
1118 	node = *rnode;
1119 	node.sysctl_data = &newtrig;
1120 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1121 	if (error || newp == NULL)
1122 		return (error);
1123 
1124 	if (newtrig != 0)
1125 		panic("Panic triggered");
1126 
1127 	return (error);
1128 }
1129 #endif
1130 
1131 /*
1132  * sysctl helper routine for kern.maxvnodes.  Drain vnodes if
1133  * new value is lower than desiredvnodes and then calls reinit
1134  * routines that needs to adjust to the new value.
1135  */
1136 static int
1137 sysctl_kern_maxvnodes(SYSCTLFN_ARGS)
1138 {
1139 	int error, new_vnodes, old_vnodes, new_max;
1140 	struct sysctlnode node;
1141 
1142 	new_vnodes = desiredvnodes;
1143 	node = *rnode;
1144 	node.sysctl_data = &new_vnodes;
1145 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1146 	if (error || newp == NULL)
1147 		return (error);
1148 
1149 	/* Limits: 75% of KVA and physical memory. */
1150 	new_max = calc_cache_size(kernel_map, 75, 75) / VNODE_COST;
1151 	if (new_vnodes > new_max)
1152 		new_vnodes = new_max;
1153 
1154 	old_vnodes = desiredvnodes;
1155 	desiredvnodes = new_vnodes;
1156 	if (new_vnodes < old_vnodes) {
1157 		error = vfs_drainvnodes(new_vnodes, l);
1158 		if (error) {
1159 			desiredvnodes = old_vnodes;
1160 			return (error);
1161 		}
1162 	}
1163 	vfs_reinit();
1164 	nchreinit();
1165 
1166 	return (0);
1167 }
1168 
1169 /*
1170  * sysctl helper routine for rtc_offset - set time after changes
1171  */
1172 static int
1173 sysctl_kern_rtc_offset(SYSCTLFN_ARGS)
1174 {
1175 	struct timespec ts, delta;
1176 	int error, new_rtc_offset;
1177 	struct sysctlnode node;
1178 
1179 	new_rtc_offset = rtc_offset;
1180 	node = *rnode;
1181 	node.sysctl_data = &new_rtc_offset;
1182 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1183 	if (error || newp == NULL)
1184 		return (error);
1185 
1186 	if (kauth_authorize_system(l->l_cred, KAUTH_SYSTEM_TIME,
1187 	    KAUTH_REQ_SYSTEM_TIME_RTCOFFSET,
1188 	    KAUTH_ARG(new_rtc_offset), NULL, NULL))
1189 		return (EPERM);
1190 	if (rtc_offset == new_rtc_offset)
1191 		return (0);
1192 
1193 	/* if we change the offset, adjust the time */
1194 	nanotime(&ts);
1195 	delta.tv_sec = 60 * (new_rtc_offset - rtc_offset);
1196 	delta.tv_nsec = 0;
1197 	timespecadd(&ts, &delta, &ts);
1198 	rtc_offset = new_rtc_offset;
1199 	return (settime(l->l_proc, &ts));
1200 }
1201 
1202 /*
1203  * sysctl helper routine for kern.maxproc. Ensures that the new
1204  * values are not too low or too high.
1205  */
1206 static int
1207 sysctl_kern_maxproc(SYSCTLFN_ARGS)
1208 {
1209 	int error, nmaxproc;
1210 	struct sysctlnode node;
1211 
1212 	nmaxproc = maxproc;
1213 	node = *rnode;
1214 	node.sysctl_data = &nmaxproc;
1215 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1216 	if (error || newp == NULL)
1217 		return (error);
1218 
1219 	if (nmaxproc < 0 || nmaxproc >= PID_MAX)
1220 		return (EINVAL);
1221 #ifdef __HAVE_CPU_MAXPROC
1222 	if (nmaxproc > cpu_maxproc())
1223 		return (EINVAL);
1224 #endif
1225 	maxproc = nmaxproc;
1226 
1227 	return (0);
1228 }
1229 
1230 /*
1231  * sysctl helper function for kern.hostid. The hostid is a long, but
1232  * we export it as an int, so we need to give it a little help.
1233  */
1234 static int
1235 sysctl_kern_hostid(SYSCTLFN_ARGS)
1236 {
1237 	int error, inthostid;
1238 	struct sysctlnode node;
1239 
1240 	inthostid = hostid;  /* XXX assumes sizeof int <= sizeof long */
1241 	node = *rnode;
1242 	node.sysctl_data = &inthostid;
1243 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1244 	if (error || newp == NULL)
1245 		return (error);
1246 
1247 	hostid = (unsigned)inthostid;
1248 
1249 	return (0);
1250 }
1251 
1252 /*
1253  * sysctl helper function for kern.hostname and kern.domainnname.
1254  * resets the relevant recorded length when the underlying name is
1255  * changed.
1256  */
1257 static int
1258 sysctl_setlen(SYSCTLFN_ARGS)
1259 {
1260 	int error;
1261 
1262 	error = sysctl_lookup(SYSCTLFN_CALL(rnode));
1263 	if (error || newp == NULL)
1264 		return (error);
1265 
1266 	switch (rnode->sysctl_num) {
1267 	case KERN_HOSTNAME:
1268 		hostnamelen = strlen((const char*)rnode->sysctl_data);
1269 		break;
1270 	case KERN_DOMAINNAME:
1271 		domainnamelen = strlen((const char*)rnode->sysctl_data);
1272 		break;
1273 	}
1274 
1275 	return (0);
1276 }
1277 
1278 /*
1279  * sysctl helper routine for kern.clockrate. Assembles a struct on
1280  * the fly to be returned to the caller.
1281  */
1282 static int
1283 sysctl_kern_clockrate(SYSCTLFN_ARGS)
1284 {
1285 	struct clockinfo clkinfo;
1286 	struct sysctlnode node;
1287 
1288 	clkinfo.tick = tick;
1289 	clkinfo.tickadj = tickadj;
1290 	clkinfo.hz = hz;
1291 	clkinfo.profhz = profhz;
1292 	clkinfo.stathz = stathz ? stathz : hz;
1293 
1294 	node = *rnode;
1295 	node.sysctl_data = &clkinfo;
1296 	return (sysctl_lookup(SYSCTLFN_CALL(&node)));
1297 }
1298 
1299 
1300 /*
1301  * sysctl helper routine for kern.file pseudo-subtree.
1302  */
1303 static int
1304 sysctl_kern_file(SYSCTLFN_ARGS)
1305 {
1306 	int error;
1307 	size_t buflen;
1308 	struct file *fp, *dp, *np, fbuf;
1309 	char *start, *where;
1310 
1311 	start = where = oldp;
1312 	buflen = *oldlenp;
1313 	dp = NULL;
1314 
1315 	if (where == NULL) {
1316 		/*
1317 		 * overestimate by 10 files
1318 		 */
1319 		*oldlenp = sizeof(filehead) + (nfiles + 10) *
1320 		    sizeof(struct file);
1321 		return (0);
1322 	}
1323 
1324 	/*
1325 	 * first dcopyout filehead
1326 	 */
1327 	if (buflen < sizeof(filehead)) {
1328 		*oldlenp = 0;
1329 		return (0);
1330 	}
1331 	sysctl_unlock();
1332 	error = dcopyout(l, &filehead, where, sizeof(filehead));
1333 	if (error) {
1334 	 	sysctl_relock();
1335 		return error;
1336 	}
1337 	buflen -= sizeof(filehead);
1338 	where += sizeof(filehead);
1339 
1340 	/*
1341 	 * allocate dummy file descriptor to make position in list
1342 	 */
1343 	if ((dp = fgetdummy()) == NULL) {
1344 	 	sysctl_relock();
1345 		return ENOMEM;
1346 	}
1347 
1348 	/*
1349 	 * followed by an array of file structures
1350 	 */
1351 	mutex_enter(&filelist_lock);
1352 	for (fp = LIST_FIRST(&filehead); fp != NULL; fp = np) {
1353 	    	np = LIST_NEXT(fp, f_list);
1354 	    	mutex_enter(&fp->f_lock);
1355 	    	if (fp->f_count == 0) {
1356 		    	mutex_exit(&fp->f_lock);
1357 	    		continue;
1358 		}
1359 		/*
1360 		 * XXX Need to prevent that from being an alternative way
1361 		 * XXX to getting process information.
1362 		 */
1363 		if (kauth_authorize_generic(l->l_cred,
1364 		    KAUTH_GENERIC_CANSEE, fp->f_cred) != 0) {
1365 		    	mutex_exit(&fp->f_lock);
1366 			continue;
1367 		}
1368 		if (buflen < sizeof(struct file)) {
1369 			*oldlenp = where - start;
1370 		    	mutex_exit(&fp->f_lock);
1371 			error = ENOMEM;
1372 			break;
1373 		}
1374 		memcpy(&fbuf, fp, sizeof(fbuf));
1375 		LIST_INSERT_AFTER(fp, dp, f_list);
1376 	    	mutex_exit(&fp->f_lock);
1377 		mutex_exit(&filelist_lock);
1378 		error = dcopyout(l, &fbuf, where, sizeof(fbuf));
1379 		if (error) {
1380 			mutex_enter(&filelist_lock);
1381 			LIST_REMOVE(dp, f_list);
1382 			break;
1383 		}
1384 		buflen -= sizeof(struct file);
1385 		where += sizeof(struct file);
1386 		mutex_enter(&filelist_lock);
1387 		np = LIST_NEXT(dp, f_list);
1388 		LIST_REMOVE(dp, f_list);
1389 	}
1390 	mutex_exit(&filelist_lock);
1391 	*oldlenp = where - start;
1392  	if (dp != NULL)
1393 		fputdummy(dp);
1394  	sysctl_relock();
1395 	return (error);
1396 }
1397 
1398 /*
1399  * sysctl helper routine for kern.msgbufsize and kern.msgbuf. For the
1400  * former it merely checks the message buffer is set up. For the latter,
1401  * it also copies out the data if necessary.
1402  */
1403 static int
1404 sysctl_msgbuf(SYSCTLFN_ARGS)
1405 {
1406 	char *where = oldp;
1407 	size_t len, maxlen;
1408 	long beg, end;
1409 	extern kmutex_t log_lock;
1410 	int error;
1411 
1412 	if (!msgbufenabled || msgbufp->msg_magic != MSG_MAGIC) {
1413 		msgbufenabled = 0;
1414 		return (ENXIO);
1415 	}
1416 
1417 	switch (rnode->sysctl_num) {
1418 	case KERN_MSGBUFSIZE: {
1419 		struct sysctlnode node = *rnode;
1420 		int msg_bufs = (int)msgbufp->msg_bufs;
1421 		node.sysctl_data = &msg_bufs;
1422 		return (sysctl_lookup(SYSCTLFN_CALL(&node)));
1423 	}
1424 	case KERN_MSGBUF:
1425 		break;
1426 	default:
1427 		return (EOPNOTSUPP);
1428 	}
1429 
1430 	if (newp != NULL)
1431 		return (EPERM);
1432 
1433 	if (oldp == NULL) {
1434 		/* always return full buffer size */
1435 		*oldlenp = msgbufp->msg_bufs;
1436 		return (0);
1437 	}
1438 
1439 	sysctl_unlock();
1440 
1441 	/*
1442 	 * First, copy from the write pointer to the end of
1443 	 * message buffer.
1444 	 */
1445 	error = 0;
1446 	mutex_spin_enter(&log_lock);
1447 	maxlen = MIN(msgbufp->msg_bufs, *oldlenp);
1448 	beg = msgbufp->msg_bufx;
1449 	end = msgbufp->msg_bufs;
1450 	mutex_spin_exit(&log_lock);
1451 
1452 	while (maxlen > 0) {
1453 		len = MIN(end - beg, maxlen);
1454 		if (len == 0)
1455 			break;
1456 		/* XXX unlocked, but hardly matters. */
1457 		error = dcopyout(l, &msgbufp->msg_bufc[beg], where, len);
1458 		if (error)
1459 			break;
1460 		where += len;
1461 		maxlen -= len;
1462 
1463 		/*
1464 		 * ... then, copy from the beginning of message buffer to
1465 		 * the write pointer.
1466 		 */
1467 		beg = 0;
1468 		end = msgbufp->msg_bufx;
1469 	}
1470 
1471 	sysctl_relock();
1472 	return (error);
1473 }
1474 
1475 /*
1476  * sysctl helper routine for kern.defcorename. In the case of a new
1477  * string being assigned, check that it's not a zero-length string.
1478  * (XXX the check in -current doesn't work, but do we really care?)
1479  */
1480 static int
1481 sysctl_kern_defcorename(SYSCTLFN_ARGS)
1482 {
1483 	int error;
1484 	char *newcorename;
1485 	struct sysctlnode node;
1486 
1487 	newcorename = PNBUF_GET();
1488 	node = *rnode;
1489 	node.sysctl_data = &newcorename[0];
1490 	memcpy(node.sysctl_data, rnode->sysctl_data, MAXPATHLEN);
1491 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1492 	if (error || newp == NULL) {
1493 		goto done;
1494 	}
1495 
1496 	/*
1497 	 * when sysctl_lookup() deals with a string, it's guaranteed
1498 	 * to come back nul terminated. So there.  :)
1499 	 */
1500 	if (strlen(newcorename) == 0) {
1501 		error = EINVAL;
1502 	} else {
1503 		memcpy(rnode->sysctl_data, node.sysctl_data, MAXPATHLEN);
1504 		error = 0;
1505 	}
1506 done:
1507 	PNBUF_PUT(newcorename);
1508 	return error;
1509 }
1510 
1511 /*
1512  * sysctl helper routine for kern.cp_time node. Adds up cpu time
1513  * across all cpus.
1514  */
1515 static int
1516 sysctl_kern_cptime(SYSCTLFN_ARGS)
1517 {
1518 	struct sysctlnode node = *rnode;
1519 	uint64_t *cp_time = NULL;
1520 	int error, n = ncpu, i;
1521 	struct cpu_info *ci;
1522 	CPU_INFO_ITERATOR cii;
1523 
1524 	/*
1525 	 * if you specifically pass a buffer that is the size of the
1526 	 * sum, or if you are probing for the size, you get the "sum"
1527 	 * of cp_time (and the size thereof) across all processors.
1528 	 *
1529 	 * alternately, you can pass an additional mib number and get
1530 	 * cp_time for that particular processor.
1531 	 */
1532 	switch (namelen) {
1533 	case 0:
1534 		if (*oldlenp == sizeof(uint64_t) * CPUSTATES || oldp == NULL) {
1535 			node.sysctl_size = sizeof(uint64_t) * CPUSTATES;
1536 			n = -1; /* SUM */
1537 		}
1538 		else {
1539 			node.sysctl_size = n * sizeof(uint64_t) * CPUSTATES;
1540 			n = -2; /* ALL */
1541 		}
1542 		break;
1543 	case 1:
1544 		if (name[0] < 0 || name[0] >= n)
1545 			return (ENOENT); /* ENOSUCHPROCESSOR */
1546 		node.sysctl_size = sizeof(uint64_t) * CPUSTATES;
1547 		n = name[0];
1548 		/*
1549 		 * adjust these so that sysctl_lookup() will be happy
1550 		 */
1551 		name++;
1552 		namelen--;
1553 		break;
1554 	default:
1555 		return (EINVAL);
1556 	}
1557 
1558 	cp_time = kmem_alloc(node.sysctl_size, KM_SLEEP);
1559 	if (cp_time == NULL)
1560 		return (ENOMEM);
1561 	node.sysctl_data = cp_time;
1562 	memset(cp_time, 0, node.sysctl_size);
1563 
1564 	for (CPU_INFO_FOREACH(cii, ci)) {
1565 		if (n <= 0) {
1566 			for (i = 0; i < CPUSTATES; i++) {
1567 				cp_time[i] += ci->ci_schedstate.spc_cp_time[i];
1568 			}
1569 		}
1570 		/*
1571 		 * if a specific processor was requested and we just
1572 		 * did it, we're done here
1573 		 */
1574 		if (n == 0)
1575 			break;
1576 		/*
1577 		 * if doing "all", skip to next cp_time set for next processor
1578 		 */
1579 		if (n == -2)
1580 			cp_time += CPUSTATES;
1581 		/*
1582 		 * if we're doing a specific processor, we're one
1583 		 * processor closer
1584 		 */
1585 		if (n > 0)
1586 			n--;
1587 	}
1588 
1589 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1590 	kmem_free(node.sysctl_data, node.sysctl_size);
1591 	return (error);
1592 }
1593 
1594 #if NPTY > 0
1595 /*
1596  * sysctl helper routine for kern.maxptys. Ensures that any new value
1597  * is acceptable to the pty subsystem.
1598  */
1599 static int
1600 sysctl_kern_maxptys(SYSCTLFN_ARGS)
1601 {
1602 	int pty_maxptys(int, int);		/* defined in kern/tty_pty.c */
1603 	int error, xmax;
1604 	struct sysctlnode node;
1605 
1606 	/* get current value of maxptys */
1607 	xmax = pty_maxptys(0, 0);
1608 
1609 	node = *rnode;
1610 	node.sysctl_data = &xmax;
1611 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1612 	if (error || newp == NULL)
1613 		return (error);
1614 
1615 	if (xmax != pty_maxptys(xmax, 1))
1616 		return (EINVAL);
1617 
1618 	return (0);
1619 }
1620 #endif /* NPTY > 0 */
1621 
1622 /*
1623  * sysctl helper routine for kern.sbmax. Basically just ensures that
1624  * any new value is not too small.
1625  */
1626 static int
1627 sysctl_kern_sbmax(SYSCTLFN_ARGS)
1628 {
1629 	int error, new_sbmax;
1630 	struct sysctlnode node;
1631 
1632 	new_sbmax = sb_max;
1633 	node = *rnode;
1634 	node.sysctl_data = &new_sbmax;
1635 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1636 	if (error || newp == NULL)
1637 		return (error);
1638 
1639 	KERNEL_LOCK(1, NULL);
1640 	error = sb_max_set(new_sbmax);
1641 	KERNEL_UNLOCK_ONE(NULL);
1642 
1643 	return (error);
1644 }
1645 
1646 /*
1647  * sysctl helper routine for kern.urandom node. Picks a random number
1648  * for you.
1649  */
1650 static int
1651 sysctl_kern_urnd(SYSCTLFN_ARGS)
1652 {
1653 #if NRND > 0
1654 	int v, rv;
1655 
1656 	KERNEL_LOCK(1, NULL);
1657 	rv = rnd_extract_data(&v, sizeof(v), RND_EXTRACT_ANY);
1658 	KERNEL_UNLOCK_ONE(NULL);
1659 	if (rv == sizeof(v)) {
1660 		struct sysctlnode node = *rnode;
1661 		node.sysctl_data = &v;
1662 		return (sysctl_lookup(SYSCTLFN_CALL(&node)));
1663 	}
1664 	else
1665 		return (EIO);	/*XXX*/
1666 #else
1667 	return (EOPNOTSUPP);
1668 #endif
1669 }
1670 
1671 /*
1672  * sysctl helper routine for kern.arandom node. Picks a random number
1673  * for you.
1674  */
1675 static int
1676 sysctl_kern_arnd(SYSCTLFN_ARGS)
1677 {
1678 #if NRND > 0
1679 	int error;
1680 	void *v;
1681 	struct sysctlnode node = *rnode;
1682 
1683 	if (*oldlenp == 0)
1684 		return 0;
1685 	if (*oldlenp > 8192)
1686 		return E2BIG;
1687 
1688 	v = kmem_alloc(*oldlenp, KM_SLEEP);
1689 	arc4randbytes(v, *oldlenp);
1690 	node.sysctl_data = v;
1691 	node.sysctl_size = *oldlenp;
1692 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1693 	kmem_free(v, *oldlenp);
1694 	return error;
1695 #else
1696 	return (EOPNOTSUPP);
1697 #endif
1698 }
1699 /*
1700  * sysctl helper routine to do kern.lwp.* work.
1701  */
1702 static int
1703 sysctl_kern_lwp(SYSCTLFN_ARGS)
1704 {
1705 	struct kinfo_lwp klwp;
1706 	struct proc *p;
1707 	struct lwp *l2, *l3;
1708 	char *where, *dp;
1709 	int pid, elem_size, elem_count;
1710 	int buflen, needed, error;
1711 	bool gotit;
1712 
1713 	if (namelen == 1 && name[0] == CTL_QUERY)
1714 		return (sysctl_query(SYSCTLFN_CALL(rnode)));
1715 
1716 	dp = where = oldp;
1717 	buflen = where != NULL ? *oldlenp : 0;
1718 	error = needed = 0;
1719 
1720 	if (newp != NULL || namelen != 3)
1721 		return (EINVAL);
1722 	pid = name[0];
1723 	elem_size = name[1];
1724 	elem_count = name[2];
1725 
1726 	sysctl_unlock();
1727 	if (pid == -1) {
1728 		mutex_enter(proc_lock);
1729 		LIST_FOREACH(p, &allproc, p_list) {
1730 			/* Grab a hold on the process. */
1731 			if (!rw_tryenter(&p->p_reflock, RW_READER)) {
1732 				continue;
1733 			}
1734 			mutex_exit(proc_lock);
1735 
1736 			mutex_enter(p->p_lock);
1737 			LIST_FOREACH(l2, &p->p_lwps, l_sibling) {
1738 				if (buflen >= elem_size && elem_count > 0) {
1739 					lwp_lock(l2);
1740 					fill_lwp(l2, &klwp);
1741 					lwp_unlock(l2);
1742 					mutex_exit(p->p_lock);
1743 
1744 					/*
1745 					 * Copy out elem_size, but not
1746 					 * larger than the size of a
1747 					 * struct kinfo_proc2.
1748 					 */
1749 					error = dcopyout(l, &klwp, dp,
1750 					    min(sizeof(klwp), elem_size));
1751 					if (error) {
1752 						rw_exit(&p->p_reflock);
1753 						goto cleanup;
1754 					}
1755 					mutex_enter(p->p_lock);
1756 					LIST_FOREACH(l3, &p->p_lwps,
1757 					    l_sibling) {
1758 						if (l2 == l3)
1759 							break;
1760 					}
1761 					if (l3 == NULL) {
1762 						mutex_exit(p->p_lock);
1763 						rw_exit(&p->p_reflock);
1764 						error = EAGAIN;
1765 						goto cleanup;
1766 					}
1767 					dp += elem_size;
1768 					buflen -= elem_size;
1769 					elem_count--;
1770 				}
1771 				needed += elem_size;
1772 			}
1773 			mutex_exit(p->p_lock);
1774 
1775 			/* Drop reference to process. */
1776 			mutex_enter(proc_lock);
1777 			rw_exit(&p->p_reflock);
1778 		}
1779 		mutex_exit(proc_lock);
1780 	} else {
1781 		mutex_enter(proc_lock);
1782 		p = p_find(pid, PFIND_LOCKED);
1783 		if (p == NULL) {
1784 			error = ESRCH;
1785 			mutex_exit(proc_lock);
1786 			goto cleanup;
1787 		}
1788 		/* Grab a hold on the process. */
1789 		gotit = rw_tryenter(&p->p_reflock, RW_READER);
1790 		mutex_exit(proc_lock);
1791 		if (!gotit) {
1792 			error = ESRCH;
1793 			goto cleanup;
1794 		}
1795 
1796 		mutex_enter(p->p_lock);
1797 		LIST_FOREACH(l2, &p->p_lwps, l_sibling) {
1798 			if (buflen >= elem_size && elem_count > 0) {
1799 				lwp_lock(l2);
1800 				fill_lwp(l2, &klwp);
1801 				lwp_unlock(l2);
1802 				mutex_exit(p->p_lock);
1803 				/*
1804 				 * Copy out elem_size, but not larger than
1805 				 * the size of a struct kinfo_proc2.
1806 				 */
1807 				error = dcopyout(l, &klwp, dp,
1808 				    min(sizeof(klwp), elem_size));
1809 				if (error) {
1810 					rw_exit(&p->p_reflock);
1811 					goto cleanup;
1812 				}
1813 				mutex_enter(p->p_lock);
1814 				LIST_FOREACH(l3, &p->p_lwps, l_sibling) {
1815 					if (l2 == l3)
1816 						break;
1817 				}
1818 				if (l3 == NULL) {
1819 					mutex_exit(p->p_lock);
1820 					rw_exit(&p->p_reflock);
1821 					error = EAGAIN;
1822 					goto cleanup;
1823 				}
1824 				dp += elem_size;
1825 				buflen -= elem_size;
1826 				elem_count--;
1827 			}
1828 			needed += elem_size;
1829 		}
1830 		mutex_exit(p->p_lock);
1831 
1832 		/* Drop reference to process. */
1833 		rw_exit(&p->p_reflock);
1834 	}
1835 
1836 	if (where != NULL) {
1837 		*oldlenp = dp - where;
1838 		if (needed > *oldlenp) {
1839 			sysctl_relock();
1840 			return (ENOMEM);
1841 		}
1842 	} else {
1843 		needed += KERN_LWPSLOP;
1844 		*oldlenp = needed;
1845 	}
1846 	error = 0;
1847  cleanup:
1848 	sysctl_relock();
1849 	return (error);
1850 }
1851 
1852 /*
1853  * sysctl helper routine for kern.forkfsleep node. Ensures that the
1854  * given value is not too large or two small, and is at least one
1855  * timer tick if not zero.
1856  */
1857 static int
1858 sysctl_kern_forkfsleep(SYSCTLFN_ARGS)
1859 {
1860 	/* userland sees value in ms, internally is in ticks */
1861 	extern int forkfsleep;		/* defined in kern/kern_fork.c */
1862 	int error, timo, lsleep;
1863 	struct sysctlnode node;
1864 
1865 	lsleep = forkfsleep * 1000 / hz;
1866 	node = *rnode;
1867 	node.sysctl_data = &lsleep;
1868 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1869 	if (error || newp == NULL)
1870 		return (error);
1871 
1872 	/* refuse negative values, and overly 'long time' */
1873 	if (lsleep < 0 || lsleep > MAXSLP * 1000)
1874 		return (EINVAL);
1875 
1876 	timo = mstohz(lsleep);
1877 
1878 	/* if the interval is >0 ms && <1 tick, use 1 tick */
1879 	if (lsleep != 0 && timo == 0)
1880 		forkfsleep = 1;
1881 	else
1882 		forkfsleep = timo;
1883 
1884 	return (0);
1885 }
1886 
1887 /*
1888  * sysctl helper routine for kern.root_partition
1889  */
1890 static int
1891 sysctl_kern_root_partition(SYSCTLFN_ARGS)
1892 {
1893 	int rootpart = DISKPART(rootdev);
1894 	struct sysctlnode node = *rnode;
1895 
1896 	node.sysctl_data = &rootpart;
1897 	return (sysctl_lookup(SYSCTLFN_CALL(&node)));
1898 }
1899 
1900 /*
1901  * sysctl helper function for kern.drivers
1902  */
1903 static int
1904 sysctl_kern_drivers(SYSCTLFN_ARGS)
1905 {
1906 	int error;
1907 	size_t buflen;
1908 	struct kinfo_drivers kd;
1909 	char *start, *where;
1910 	const char *dname;
1911 	int i;
1912 	extern struct devsw_conv *devsw_conv;
1913 	extern int max_devsw_convs;
1914 
1915 	if (newp != NULL || namelen != 0)
1916 		return (EINVAL);
1917 
1918 	start = where = oldp;
1919 	buflen = *oldlenp;
1920 	if (where == NULL) {
1921 		*oldlenp = max_devsw_convs * sizeof kd;
1922 		return 0;
1923 	}
1924 
1925 	/*
1926 	 * An array of kinfo_drivers structures
1927 	 */
1928 	error = 0;
1929 	sysctl_unlock();
1930 	mutex_enter(&device_lock);
1931 	for (i = 0; i < max_devsw_convs; i++) {
1932 		dname = devsw_conv[i].d_name;
1933 		if (dname == NULL)
1934 			continue;
1935 		if (buflen < sizeof kd) {
1936 			error = ENOMEM;
1937 			break;
1938 		}
1939 		memset(&kd, 0, sizeof(kd));
1940 		kd.d_bmajor = devsw_conv[i].d_bmajor;
1941 		kd.d_cmajor = devsw_conv[i].d_cmajor;
1942 		strlcpy(kd.d_name, dname, sizeof kd.d_name);
1943 		mutex_exit(&device_lock);
1944 		error = dcopyout(l, &kd, where, sizeof kd);
1945 		mutex_enter(&device_lock);
1946 		if (error != 0)
1947 			break;
1948 		buflen -= sizeof kd;
1949 		where += sizeof kd;
1950 	}
1951 	mutex_exit(&device_lock);
1952 	sysctl_relock();
1953 	*oldlenp = where - start;
1954 	return error;
1955 }
1956 
1957 /*
1958  * sysctl helper function for kern.file2
1959  */
1960 static int
1961 sysctl_kern_file2(SYSCTLFN_ARGS)
1962 {
1963 	struct proc *p;
1964 	struct file *fp, *tp, *np;
1965 	struct filedesc *fd;
1966 	struct kinfo_file kf;
1967 	char *dp;
1968 	u_int i, op;
1969 	size_t len, needed, elem_size, out_size;
1970 	int error, arg, elem_count;
1971 	fdfile_t *ff;
1972 
1973 	if (namelen == 1 && name[0] == CTL_QUERY)
1974 		return (sysctl_query(SYSCTLFN_CALL(rnode)));
1975 
1976 	if (namelen != 4)
1977 		return (EINVAL);
1978 
1979 	error = 0;
1980 	dp = oldp;
1981 	len = (oldp != NULL) ? *oldlenp : 0;
1982 	op = name[0];
1983 	arg = name[1];
1984 	elem_size = name[2];
1985 	elem_count = name[3];
1986 	out_size = MIN(sizeof(kf), elem_size);
1987 	needed = 0;
1988 
1989 	if (elem_size < 1 || elem_count < 0)
1990 		return (EINVAL);
1991 
1992 	switch (op) {
1993 	case KERN_FILE_BYFILE:
1994 		/*
1995 		 * doesn't use arg so it must be zero
1996 		 */
1997 		if (arg != 0)
1998 			return (EINVAL);
1999 		sysctl_unlock();
2000 		/*
2001 		 * allocate dummy file descriptor to make position in list
2002 		 */
2003 		if ((tp = fgetdummy()) == NULL) {
2004 		 	sysctl_relock();
2005 			return ENOMEM;
2006 		}
2007 		mutex_enter(&filelist_lock);
2008 		for (fp = LIST_FIRST(&filehead); fp != NULL; fp = np) {
2009 			np = LIST_NEXT(fp, f_list);
2010 			mutex_enter(&fp->f_lock);
2011 			if (fp->f_count == 0) {
2012 				mutex_exit(&fp->f_lock);
2013 				continue;
2014 			}
2015 			/*
2016 			 * XXX Need to prevent that from being an alternative
2017 			 * XXX way for getting process information.
2018 			 */
2019 			if (kauth_authorize_generic(l->l_cred,
2020 			    KAUTH_GENERIC_CANSEE, fp->f_cred) != 0) {
2021 				mutex_exit(&fp->f_lock);
2022 				continue;
2023 			}
2024 			if (len >= elem_size && elem_count > 0) {
2025 				fill_file(&kf, fp, NULL, 0, 0);
2026 				LIST_INSERT_AFTER(fp, tp, f_list);
2027 				mutex_exit(&fp->f_lock);
2028 				mutex_exit(&filelist_lock);
2029 				error = dcopyout(l, &kf, dp, out_size);
2030 				mutex_enter(&filelist_lock);
2031 				np = LIST_NEXT(tp, f_list);
2032 				LIST_REMOVE(tp, f_list);
2033 				if (error) {
2034 					break;
2035 				}
2036 				dp += elem_size;
2037 				len -= elem_size;
2038 			} else {
2039 				mutex_exit(&fp->f_lock);
2040 			}
2041 			if (elem_count > 0) {
2042 				needed += elem_size;
2043 				if (elem_count != INT_MAX)
2044 					elem_count--;
2045 			}
2046 		}
2047 		mutex_exit(&filelist_lock);
2048 		fputdummy(tp);
2049 		sysctl_relock();
2050 		break;
2051 	case KERN_FILE_BYPID:
2052 		if (arg < -1)
2053 			/* -1 means all processes */
2054 			return (EINVAL);
2055 		sysctl_unlock();
2056 		mutex_enter(proc_lock);
2057 		LIST_FOREACH(p, &allproc, p_list) {
2058 			if (p->p_stat == SIDL) {
2059 				/* skip embryonic processes */
2060 				continue;
2061 			}
2062 			if (arg > 0 && p->p_pid != arg) {
2063 				/* pick only the one we want */
2064 				/* XXX want 0 to mean "kernel files" */
2065 				continue;
2066 			}
2067 			mutex_enter(p->p_lock);
2068 			error = kauth_authorize_process(l->l_cred,
2069 			    KAUTH_PROCESS_CANSEE, p,
2070 			    KAUTH_ARG(KAUTH_REQ_PROCESS_CANSEE_OPENFILES),
2071 			    NULL, NULL);
2072 			mutex_exit(p->p_lock);
2073 			if (error != 0) {
2074 				/*
2075 				 * Don't leak kauth retval if we're silently
2076 				 * skipping this entry.
2077 				 */
2078 				error = 0;
2079 				continue;
2080 			}
2081 
2082 			/*
2083 			 * Grab a hold on the process.
2084 			 */
2085 			if (!rw_tryenter(&p->p_reflock, RW_READER)) {
2086 				continue;
2087 			}
2088 			mutex_exit(proc_lock);
2089 
2090 			/* XXX Do we need to check permission per file? */
2091 			fd = p->p_fd;
2092 			mutex_enter(&fd->fd_lock);
2093 			for (i = 0; i < fd->fd_nfiles; i++) {
2094 				if ((ff = fd->fd_ofiles[i]) == NULL) {
2095 					continue;
2096 				}
2097 				mutex_enter(&ff->ff_lock);
2098 				if ((fp = ff->ff_file) == NULL) {
2099 					mutex_exit(&ff->ff_lock);
2100 					continue;
2101 				}
2102 				if (len >= elem_size && elem_count > 0) {
2103 					mutex_enter(&fp->f_lock);
2104 					fill_file(&kf, fp, ff, i, p->p_pid);
2105 					mutex_exit(&fp->f_lock);
2106 					mutex_exit(&ff->ff_lock);
2107 					mutex_exit(&fd->fd_lock);
2108 					error = dcopyout(l, &kf, dp, out_size);
2109 					mutex_enter(&fd->fd_lock);
2110 					if (error)
2111 						break;
2112 					dp += elem_size;
2113 					len -= elem_size;
2114 				} else {
2115 					mutex_exit(&ff->ff_lock);
2116 				}
2117 				if (elem_count > 0) {
2118 					needed += elem_size;
2119 					if (elem_count != INT_MAX)
2120 						elem_count--;
2121 				}
2122 			}
2123 			mutex_exit(&fd->fd_lock);
2124 
2125 			/*
2126 			 * Release reference to process.
2127 			 */
2128 			mutex_enter(proc_lock);
2129 			rw_exit(&p->p_reflock);
2130 		}
2131 		mutex_exit(proc_lock);
2132 		sysctl_relock();
2133 		break;
2134 	default:
2135 		return (EINVAL);
2136 	}
2137 
2138 	if (oldp == NULL)
2139 		needed += KERN_FILESLOP * elem_size;
2140 	*oldlenp = needed;
2141 
2142 	return (error);
2143 }
2144 
2145 static void
2146 fill_file(struct kinfo_file *kp, const file_t *fp, const fdfile_t *ff,
2147 	  int i, pid_t pid)
2148 {
2149 
2150 	memset(kp, 0, sizeof(*kp));
2151 
2152 	kp->ki_fileaddr =	PTRTOUINT64(fp);
2153 	kp->ki_flag =		fp->f_flag;
2154 	kp->ki_iflags =		fp->f_iflags;
2155 	kp->ki_ftype =		fp->f_type;
2156 	kp->ki_count =		fp->f_count;
2157 	kp->ki_msgcount =	fp->f_msgcount;
2158 	kp->ki_fucred =		PTRTOUINT64(fp->f_cred);
2159 	kp->ki_fuid =		kauth_cred_geteuid(fp->f_cred);
2160 	kp->ki_fgid =		kauth_cred_getegid(fp->f_cred);
2161 	kp->ki_fops =		PTRTOUINT64(fp->f_ops);
2162 	kp->ki_foffset =	fp->f_offset;
2163 	kp->ki_fdata =		PTRTOUINT64(fp->f_data);
2164 
2165 	/* vnode information to glue this file to something */
2166 	if (fp->f_type == DTYPE_VNODE) {
2167 		struct vnode *vp = (struct vnode *)fp->f_data;
2168 
2169 		kp->ki_vun =	PTRTOUINT64(vp->v_un.vu_socket);
2170 		kp->ki_vsize =	vp->v_size;
2171 		kp->ki_vtype =	vp->v_type;
2172 		kp->ki_vtag =	vp->v_tag;
2173 		kp->ki_vdata =	PTRTOUINT64(vp->v_data);
2174 	}
2175 
2176 	/* process information when retrieved via KERN_FILE_BYPID */
2177 	if (ff != NULL) {
2178 		kp->ki_pid =		pid;
2179 		kp->ki_fd =		i;
2180 		kp->ki_ofileflags =	ff->ff_exclose;
2181 		kp->ki_usecount =	ff->ff_refcnt;
2182 	}
2183 }
2184 
2185 static int
2186 sysctl_doeproc(SYSCTLFN_ARGS)
2187 {
2188 	struct eproc *eproc;
2189 	struct kinfo_proc2 *kproc2;
2190 	struct kinfo_proc *dp;
2191 	struct proc *p, *next, *marker;
2192 	char *where, *dp2;
2193 	int type, op, arg, error;
2194 	u_int elem_size, elem_count;
2195 	size_t buflen, needed;
2196 	bool match, zombie, mmmbrains;
2197 
2198 	if (namelen == 1 && name[0] == CTL_QUERY)
2199 		return (sysctl_query(SYSCTLFN_CALL(rnode)));
2200 
2201 	dp = oldp;
2202 	dp2 = where = oldp;
2203 	buflen = where != NULL ? *oldlenp : 0;
2204 	error = 0;
2205 	needed = 0;
2206 	type = rnode->sysctl_num;
2207 
2208 	if (type == KERN_PROC) {
2209 		if (namelen != 2 && !(namelen == 1 && name[0] == KERN_PROC_ALL))
2210 			return (EINVAL);
2211 		op = name[0];
2212 		if (op != KERN_PROC_ALL)
2213 			arg = name[1];
2214 		else
2215 			arg = 0;		/* Quell compiler warning */
2216 		elem_size = elem_count = 0;	/* Ditto */
2217 	} else {
2218 		if (namelen != 4)
2219 			return (EINVAL);
2220 		op = name[0];
2221 		arg = name[1];
2222 		elem_size = name[2];
2223 		elem_count = name[3];
2224 	}
2225 
2226 	sysctl_unlock();
2227 
2228 	if (type == KERN_PROC) {
2229 		eproc = kmem_alloc(sizeof(*eproc), KM_SLEEP);
2230 		kproc2 = NULL;
2231 	} else {
2232 		eproc = NULL;
2233 		kproc2 = kmem_alloc(sizeof(*kproc2), KM_SLEEP);
2234 	}
2235 	marker = kmem_alloc(sizeof(*marker), KM_SLEEP);
2236 
2237 	mutex_enter(proc_lock);
2238 	mmmbrains = false;
2239 	for (p = LIST_FIRST(&allproc);; p = next) {
2240 		if (p == NULL) {
2241 			if (!mmmbrains) {
2242 				p = LIST_FIRST(&zombproc);
2243 				mmmbrains = true;
2244 			}
2245 			if (p == NULL)
2246 				break;
2247 		}
2248 		next = LIST_NEXT(p, p_list);
2249 
2250 		/*
2251 		 * Skip embryonic processes.
2252 		 */
2253 		if (p->p_stat == SIDL)
2254 			continue;
2255 
2256 		mutex_enter(p->p_lock);
2257 		error = kauth_authorize_process(l->l_cred,
2258 		    KAUTH_PROCESS_CANSEE, p,
2259 		    KAUTH_ARG(KAUTH_REQ_PROCESS_CANSEE_ENTRY), NULL, NULL);
2260 		if (error != 0) {
2261 			mutex_exit(p->p_lock);
2262 			continue;
2263 		}
2264 
2265 		/*
2266 		 * TODO - make more efficient (see notes below).
2267 		 * do by session.
2268 		 */
2269 		switch (op) {
2270 		case KERN_PROC_PID:
2271 			/* could do this with just a lookup */
2272 			match = (p->p_pid == (pid_t)arg);
2273 			break;
2274 
2275 		case KERN_PROC_PGRP:
2276 			/* could do this by traversing pgrp */
2277 			match = (p->p_pgrp->pg_id == (pid_t)arg);
2278 			break;
2279 
2280 		case KERN_PROC_SESSION:
2281 			match = (p->p_session->s_sid == (pid_t)arg);
2282 			break;
2283 
2284 		case KERN_PROC_TTY:
2285 			match = true;
2286 			if (arg == (int) KERN_PROC_TTY_REVOKE) {
2287 				if ((p->p_lflag & PL_CONTROLT) == 0 ||
2288 				    p->p_session->s_ttyp == NULL ||
2289 				    p->p_session->s_ttyvp != NULL) {
2290 				    	match = false;
2291 				}
2292 			} else if ((p->p_lflag & PL_CONTROLT) == 0 ||
2293 			    p->p_session->s_ttyp == NULL) {
2294 				if ((dev_t)arg != KERN_PROC_TTY_NODEV) {
2295 					match = false;
2296 				}
2297 			} else if (p->p_session->s_ttyp->t_dev != (dev_t)arg) {
2298 				match = false;
2299 			}
2300 			break;
2301 
2302 		case KERN_PROC_UID:
2303 			match = (kauth_cred_geteuid(p->p_cred) == (uid_t)arg);
2304 			break;
2305 
2306 		case KERN_PROC_RUID:
2307 			match = (kauth_cred_getuid(p->p_cred) == (uid_t)arg);
2308 			break;
2309 
2310 		case KERN_PROC_GID:
2311 			match = (kauth_cred_getegid(p->p_cred) == (uid_t)arg);
2312 			break;
2313 
2314 		case KERN_PROC_RGID:
2315 			match = (kauth_cred_getgid(p->p_cred) == (uid_t)arg);
2316 			break;
2317 
2318 		case KERN_PROC_ALL:
2319 			match = true;
2320 			/* allow everything */
2321 			break;
2322 
2323 		default:
2324 			error = EINVAL;
2325 			mutex_exit(p->p_lock);
2326 			goto cleanup;
2327 		}
2328 		if (!match) {
2329 			mutex_exit(p->p_lock);
2330 			continue;
2331 		}
2332 
2333 		/*
2334 		 * Grab a hold on the process.
2335 		 */
2336 		if (mmmbrains) {
2337 			zombie = true;
2338 		} else {
2339 			zombie = !rw_tryenter(&p->p_reflock, RW_READER);
2340 		}
2341 		if (zombie) {
2342 			LIST_INSERT_AFTER(p, marker, p_list);
2343 		}
2344 
2345 		if (type == KERN_PROC) {
2346 			if (buflen >= sizeof(struct kinfo_proc)) {
2347 				fill_eproc(p, eproc, zombie);
2348 				mutex_exit(p->p_lock);
2349 				mutex_exit(proc_lock);
2350 				error = dcopyout(l, p, &dp->kp_proc,
2351 				    sizeof(struct proc));
2352 				mutex_enter(proc_lock);
2353 				if (error) {
2354 					goto bah;
2355 				}
2356 				error = dcopyout(l, eproc, &dp->kp_eproc,
2357 				    sizeof(*eproc));
2358 				if (error) {
2359 					goto bah;
2360 				}
2361 				dp++;
2362 				buflen -= sizeof(struct kinfo_proc);
2363 			} else {
2364 				mutex_exit(p->p_lock);
2365 			}
2366 			needed += sizeof(struct kinfo_proc);
2367 		} else { /* KERN_PROC2 */
2368 			if (buflen >= elem_size && elem_count > 0) {
2369 				fill_kproc2(p, kproc2, zombie);
2370 				mutex_exit(p->p_lock);
2371 				mutex_exit(proc_lock);
2372 				/*
2373 				 * Copy out elem_size, but not larger than
2374 				 * the size of a struct kinfo_proc2.
2375 				 */
2376 				error = dcopyout(l, kproc2, dp2,
2377 				    min(sizeof(*kproc2), elem_size));
2378 				mutex_enter(proc_lock);
2379 				if (error) {
2380 					goto bah;
2381 				}
2382 				dp2 += elem_size;
2383 				buflen -= elem_size;
2384 				elem_count--;
2385 			} else {
2386 				mutex_exit(p->p_lock);
2387 			}
2388 			needed += elem_size;
2389 		}
2390 
2391 		/*
2392 		 * Release reference to process.
2393 		 */
2394 	 	if (zombie) {
2395 			next = LIST_NEXT(marker, p_list);
2396  			LIST_REMOVE(marker, p_list);
2397 		} else {
2398 			rw_exit(&p->p_reflock);
2399 		}
2400 	}
2401 	mutex_exit(proc_lock);
2402 
2403 	if (where != NULL) {
2404 		if (type == KERN_PROC)
2405 			*oldlenp = (char *)dp - where;
2406 		else
2407 			*oldlenp = dp2 - where;
2408 		if (needed > *oldlenp) {
2409 			error = ENOMEM;
2410 			goto out;
2411 		}
2412 	} else {
2413 		needed += KERN_PROCSLOP;
2414 		*oldlenp = needed;
2415 	}
2416 	if (kproc2)
2417 		kmem_free(kproc2, sizeof(*kproc2));
2418 	if (eproc)
2419 		kmem_free(eproc, sizeof(*eproc));
2420 	if (marker)
2421 		kmem_free(marker, sizeof(*marker));
2422 	sysctl_relock();
2423 	return 0;
2424  bah:
2425  	if (zombie)
2426  		LIST_REMOVE(marker, p_list);
2427 	else
2428 		rw_exit(&p->p_reflock);
2429  cleanup:
2430 	mutex_exit(proc_lock);
2431  out:
2432 	if (kproc2)
2433 		kmem_free(kproc2, sizeof(*kproc2));
2434 	if (eproc)
2435 		kmem_free(eproc, sizeof(*eproc));
2436 	if (marker)
2437 		kmem_free(marker, sizeof(*marker));
2438 	sysctl_relock();
2439 	return error;
2440 }
2441 
2442 /*
2443  * sysctl helper routine for kern.proc_args pseudo-subtree.
2444  */
2445 static int
2446 sysctl_kern_proc_args(SYSCTLFN_ARGS)
2447 {
2448 	struct ps_strings pss;
2449 	struct proc *p;
2450 	size_t len, i;
2451 	struct uio auio;
2452 	struct iovec aiov;
2453 	pid_t pid;
2454 	int nargv, type, error, argvlen;
2455 	char *arg;
2456 	char **argv = NULL;
2457 	char *tmp;
2458 	struct vmspace *vmspace;
2459 	vaddr_t psstr_addr;
2460 	vaddr_t offsetn;
2461 	vaddr_t offsetv;
2462 
2463 	if (namelen == 1 && name[0] == CTL_QUERY)
2464 		return (sysctl_query(SYSCTLFN_CALL(rnode)));
2465 
2466 	if (newp != NULL || namelen != 2)
2467 		return (EINVAL);
2468 	pid = name[0];
2469 	type = name[1];
2470 	argv = NULL;
2471 	argvlen = 0;
2472 
2473 	switch (type) {
2474 	case KERN_PROC_ARGV:
2475 	case KERN_PROC_NARGV:
2476 	case KERN_PROC_ENV:
2477 	case KERN_PROC_NENV:
2478 		/* ok */
2479 		break;
2480 	default:
2481 		return (EINVAL);
2482 	}
2483 
2484 	sysctl_unlock();
2485 
2486 	/* check pid */
2487 	mutex_enter(proc_lock);
2488 	if ((p = p_find(pid, PFIND_LOCKED)) == NULL) {
2489 		error = EINVAL;
2490 		goto out_locked;
2491 	}
2492 	mutex_enter(p->p_lock);
2493 
2494 	/* Check permission. */
2495 	if (type == KERN_PROC_ARGV || type == KERN_PROC_NARGV)
2496 		error = kauth_authorize_process(l->l_cred, KAUTH_PROCESS_CANSEE,
2497 		    p, KAUTH_ARG(KAUTH_REQ_PROCESS_CANSEE_ARGS), NULL, NULL);
2498 	else if (type == KERN_PROC_ENV || type == KERN_PROC_NENV)
2499 		error = kauth_authorize_process(l->l_cred, KAUTH_PROCESS_CANSEE,
2500 		    p, KAUTH_ARG(KAUTH_REQ_PROCESS_CANSEE_ENV), NULL, NULL);
2501 	else
2502 		error = EINVAL; /* XXXGCC */
2503 	if (error) {
2504 		mutex_exit(p->p_lock);
2505 		goto out_locked;
2506 	}
2507 
2508 	if (oldp == NULL) {
2509 		if (type == KERN_PROC_NARGV || type == KERN_PROC_NENV)
2510 			*oldlenp = sizeof (int);
2511 		else
2512 			*oldlenp = ARG_MAX;	/* XXX XXX XXX */
2513 		error = 0;
2514 		mutex_exit(p->p_lock);
2515 		goto out_locked;
2516 	}
2517 
2518 	/*
2519 	 * Zombies don't have a stack, so we can't read their psstrings.
2520 	 * System processes also don't have a user stack.
2521 	 */
2522 	if (P_ZOMBIE(p) || (p->p_flag & PK_SYSTEM) != 0) {
2523 		error = EINVAL;
2524 		mutex_exit(p->p_lock);
2525 		goto out_locked;
2526 	}
2527 
2528 	/*
2529 	 * Lock the process down in memory.
2530 	 */
2531 	psstr_addr = (vaddr_t)p->p_psstr;
2532 	if (type == KERN_PROC_ARGV || type == KERN_PROC_NARGV) {
2533 		offsetn = p->p_psnargv;
2534 		offsetv = p->p_psargv;
2535 	} else {
2536 		offsetn = p->p_psnenv;
2537 		offsetv = p->p_psenv;
2538 	}
2539 	vmspace = p->p_vmspace;
2540 	uvmspace_addref(vmspace);
2541 	mutex_exit(p->p_lock);
2542 	mutex_exit(proc_lock);
2543 
2544 	/*
2545 	 * Allocate a temporary buffer to hold the arguments.
2546 	 */
2547 	arg = kmem_alloc(PAGE_SIZE, KM_SLEEP);
2548 
2549 	/*
2550 	 * Read in the ps_strings structure.
2551 	 */
2552 	aiov.iov_base = &pss;
2553 	aiov.iov_len = sizeof(pss);
2554 	auio.uio_iov = &aiov;
2555 	auio.uio_iovcnt = 1;
2556 	auio.uio_offset = psstr_addr;
2557 	auio.uio_resid = sizeof(pss);
2558 	auio.uio_rw = UIO_READ;
2559 	UIO_SETUP_SYSSPACE(&auio);
2560 	error = uvm_io(&vmspace->vm_map, &auio);
2561 	if (error)
2562 		goto done;
2563 
2564 	memcpy(&nargv, (char *)&pss + offsetn, sizeof(nargv));
2565 	if (type == KERN_PROC_NARGV || type == KERN_PROC_NENV) {
2566 		error = dcopyout(l, &nargv, oldp, sizeof(nargv));
2567 		*oldlenp = sizeof(nargv);
2568 		goto done;
2569 	}
2570 	/*
2571 	 * Now read the address of the argument vector.
2572 	 */
2573 	switch (type) {
2574 	case KERN_PROC_ARGV:
2575 		/* FALLTHROUGH */
2576 	case KERN_PROC_ENV:
2577 		memcpy(&tmp, (char *)&pss + offsetv, sizeof(tmp));
2578 		break;
2579 	default:
2580 		error = EINVAL;
2581 		goto done;
2582 	}
2583 
2584 #ifdef COMPAT_NETBSD32
2585 	if (p->p_flag & PK_32)
2586 		len = sizeof(netbsd32_charp) * nargv;
2587 	else
2588 #endif
2589 		len = sizeof(char *) * nargv;
2590 
2591 	if ((argvlen = len) != 0)
2592 		argv = kmem_alloc(len, KM_SLEEP);
2593 
2594 	aiov.iov_base = argv;
2595 	aiov.iov_len = len;
2596 	auio.uio_iov = &aiov;
2597 	auio.uio_iovcnt = 1;
2598 	auio.uio_offset = (off_t)(unsigned long)tmp;
2599 	auio.uio_resid = len;
2600 	auio.uio_rw = UIO_READ;
2601 	UIO_SETUP_SYSSPACE(&auio);
2602 	error = uvm_io(&vmspace->vm_map, &auio);
2603 	if (error)
2604 		goto done;
2605 
2606 	/*
2607 	 * Now copy each string.
2608 	 */
2609 	len = 0; /* bytes written to user buffer */
2610 	for (i = 0; i < nargv; i++) {
2611 		int finished = 0;
2612 		vaddr_t base;
2613 		size_t xlen;
2614 		int j;
2615 
2616 #ifdef COMPAT_NETBSD32
2617 		if (p->p_flag & PK_32) {
2618 			netbsd32_charp *argv32;
2619 
2620 			argv32 = (netbsd32_charp *)argv;
2621 			base = (vaddr_t)NETBSD32PTR64(argv32[i]);
2622 		} else
2623 #endif
2624 			base = (vaddr_t)argv[i];
2625 
2626 		/*
2627 		 * The program has messed around with its arguments,
2628 		 * possibly deleting some, and replacing them with
2629 		 * NULL's. Treat this as the last argument and not
2630 		 * a failure.
2631 		 */
2632 		if (base == 0)
2633 			break;
2634 
2635 		while (!finished) {
2636 			xlen = PAGE_SIZE - (base & PAGE_MASK);
2637 
2638 			aiov.iov_base = arg;
2639 			aiov.iov_len = PAGE_SIZE;
2640 			auio.uio_iov = &aiov;
2641 			auio.uio_iovcnt = 1;
2642 			auio.uio_offset = base;
2643 			auio.uio_resid = xlen;
2644 			auio.uio_rw = UIO_READ;
2645 			UIO_SETUP_SYSSPACE(&auio);
2646 			error = uvm_io(&vmspace->vm_map, &auio);
2647 			if (error)
2648 				goto done;
2649 
2650 			/* Look for the end of the string */
2651 			for (j = 0; j < xlen; j++) {
2652 				if (arg[j] == '\0') {
2653 					xlen = j + 1;
2654 					finished = 1;
2655 					break;
2656 				}
2657 			}
2658 
2659 			/* Check for user buffer overflow */
2660 			if (len + xlen > *oldlenp) {
2661 				finished = 1;
2662 				if (len > *oldlenp)
2663 					xlen = 0;
2664 				else
2665 					xlen = *oldlenp - len;
2666 			}
2667 
2668 			/* Copyout the page */
2669 			error = dcopyout(l, arg, (char *)oldp + len, xlen);
2670 			if (error)
2671 				goto done;
2672 
2673 			len += xlen;
2674 			base += xlen;
2675 		}
2676 	}
2677 	*oldlenp = len;
2678 
2679 done:
2680 	if (argvlen != 0)
2681 		kmem_free(argv, argvlen);
2682 	uvmspace_free(vmspace);
2683 	kmem_free(arg, PAGE_SIZE);
2684 	sysctl_relock();
2685 	return error;
2686 
2687 out_locked:
2688 	mutex_exit(proc_lock);
2689 	sysctl_relock();
2690 	return error;
2691 }
2692 
2693 static int
2694 sysctl_security_setidcore(SYSCTLFN_ARGS)
2695 {
2696 	int newsize, error;
2697 	struct sysctlnode node;
2698 
2699 	node = *rnode;
2700 	node.sysctl_data = &newsize;
2701 	newsize = *(int *)rnode->sysctl_data;
2702 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
2703 	if (error || newp == NULL)
2704 		return error;
2705 
2706 	if (kauth_authorize_system(l->l_cred, KAUTH_SYSTEM_SETIDCORE,
2707 	    0, NULL, NULL, NULL))
2708 		return (EPERM);
2709 
2710 	*(int *)rnode->sysctl_data = newsize;
2711 
2712 	return 0;
2713 }
2714 
2715 static int
2716 sysctl_security_setidcorename(SYSCTLFN_ARGS)
2717 {
2718 	int error;
2719 	char *newsetidcorename;
2720 	struct sysctlnode node;
2721 
2722 	newsetidcorename = PNBUF_GET();
2723 	node = *rnode;
2724 	node.sysctl_data = newsetidcorename;
2725 	memcpy(node.sysctl_data, rnode->sysctl_data, MAXPATHLEN);
2726 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
2727 	if (error || newp == NULL) {
2728 		goto out;
2729 	}
2730 	if (kauth_authorize_system(l->l_cred, KAUTH_SYSTEM_SETIDCORE,
2731 	    0, NULL, NULL, NULL)) {
2732 		error = EPERM;
2733 		goto out;
2734 	}
2735 	if (strlen(newsetidcorename) == 0) {
2736 		error = EINVAL;
2737 		goto out;
2738 	}
2739 	memcpy(rnode->sysctl_data, node.sysctl_data, MAXPATHLEN);
2740 out:
2741 	PNBUF_PUT(newsetidcorename);
2742 	return error;
2743 }
2744 
2745 /*
2746  * sysctl helper routine for kern.cp_id node. Maps cpus to their
2747  * cpuids.
2748  */
2749 static int
2750 sysctl_kern_cpid(SYSCTLFN_ARGS)
2751 {
2752 	struct sysctlnode node = *rnode;
2753 	uint64_t *cp_id = NULL;
2754 	int error, n = ncpu;
2755 	struct cpu_info *ci;
2756 	CPU_INFO_ITERATOR cii;
2757 
2758 	/*
2759 	 * Here you may either retrieve a single cpu id or the whole
2760 	 * set. The size you get back when probing depends on what
2761 	 * you ask for.
2762 	 */
2763 	switch (namelen) {
2764 	case 0:
2765 		node.sysctl_size = n * sizeof(uint64_t);
2766 		n = -2; /* ALL */
2767 		break;
2768 	case 1:
2769 		if (name[0] < 0 || name[0] >= n)
2770 			return (ENOENT); /* ENOSUCHPROCESSOR */
2771 		node.sysctl_size = sizeof(uint64_t);
2772 		n = name[0];
2773 		/*
2774 		 * adjust these so that sysctl_lookup() will be happy
2775 		 */
2776 		name++;
2777 		namelen--;
2778 		break;
2779 	default:
2780 		return (EINVAL);
2781 	}
2782 
2783 	cp_id = kmem_alloc(node.sysctl_size, KM_SLEEP);
2784 	if (cp_id == NULL)
2785 		return (ENOMEM);
2786 	node.sysctl_data = cp_id;
2787 	memset(cp_id, 0, node.sysctl_size);
2788 
2789 	for (CPU_INFO_FOREACH(cii, ci)) {
2790 		if (n <= 0)
2791 			cp_id[0] = cpu_index(ci);
2792 		/*
2793 		 * if a specific processor was requested and we just
2794 		 * did it, we're done here
2795 		 */
2796 		if (n == 0)
2797 			break;
2798 		/*
2799 		 * if doing "all", skip to next cp_id slot for next processor
2800 		 */
2801 		if (n == -2)
2802 			cp_id++;
2803 		/*
2804 		 * if we're doing a specific processor, we're one
2805 		 * processor closer
2806 		 */
2807 		if (n > 0)
2808 			n--;
2809 	}
2810 
2811 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
2812 	kmem_free(node.sysctl_data, node.sysctl_size);
2813 	return (error);
2814 }
2815 
2816 /*
2817  * sysctl helper routine for hw.usermem and hw.usermem64. Values are
2818  * calculate on the fly taking into account integer overflow and the
2819  * current wired count.
2820  */
2821 static int
2822 sysctl_hw_usermem(SYSCTLFN_ARGS)
2823 {
2824 	u_int ui;
2825 	u_quad_t uq;
2826 	struct sysctlnode node;
2827 
2828 	node = *rnode;
2829 	switch (rnode->sysctl_num) {
2830 	    case HW_USERMEM:
2831 		if ((ui = physmem - uvmexp.wired) > (UINT_MAX / PAGE_SIZE))
2832 			ui = UINT_MAX;
2833 		else
2834 			ui *= PAGE_SIZE;
2835 		node.sysctl_data = &ui;
2836 		break;
2837 	case HW_USERMEM64:
2838 		uq = (u_quad_t)(physmem - uvmexp.wired) * PAGE_SIZE;
2839 		node.sysctl_data = &uq;
2840 		break;
2841 	default:
2842 		return (EINVAL);
2843 	}
2844 
2845 	return (sysctl_lookup(SYSCTLFN_CALL(&node)));
2846 }
2847 
2848 /*
2849  * sysctl helper routine for kern.cnmagic node. Pulls the old value
2850  * out, encoded, and stuffs the new value in for decoding.
2851  */
2852 static int
2853 sysctl_hw_cnmagic(SYSCTLFN_ARGS)
2854 {
2855 	char magic[CNS_LEN];
2856 	int error;
2857 	struct sysctlnode node;
2858 
2859 	if (oldp)
2860 		cn_get_magic(magic, CNS_LEN);
2861 	node = *rnode;
2862 	node.sysctl_data = &magic[0];
2863 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
2864 	if (error || newp == NULL)
2865 		return (error);
2866 
2867 	return (cn_set_magic(magic));
2868 }
2869 
2870 /*
2871  * ********************************************************************
2872  * section 3: public helper routines that are used for more than one
2873  * node
2874  * ********************************************************************
2875  */
2876 
2877 /*
2878  * sysctl helper routine for the kern.root_device node and some ports'
2879  * machdep.root_device nodes.
2880  */
2881 int
2882 sysctl_root_device(SYSCTLFN_ARGS)
2883 {
2884 	struct sysctlnode node;
2885 
2886 	node = *rnode;
2887 	node.sysctl_data = root_device->dv_xname;
2888 	node.sysctl_size = strlen(device_xname(root_device)) + 1;
2889 	return (sysctl_lookup(SYSCTLFN_CALL(&node)));
2890 }
2891 
2892 /*
2893  * sysctl helper routine for kern.consdev, dependent on the current
2894  * state of the console. Also used for machdep.console_device on some
2895  * ports.
2896  */
2897 int
2898 sysctl_consdev(SYSCTLFN_ARGS)
2899 {
2900 	dev_t consdev;
2901 	struct sysctlnode node;
2902 
2903 	if (cn_tab != NULL)
2904 		consdev = cn_tab->cn_dev;
2905 	else
2906 		consdev = NODEV;
2907 	node = *rnode;
2908 	node.sysctl_data = &consdev;
2909 	node.sysctl_size = sizeof(consdev);
2910 	return (sysctl_lookup(SYSCTLFN_CALL(&node)));
2911 }
2912 
2913 /*
2914  * ********************************************************************
2915  * section 4: support for some helpers
2916  * ********************************************************************
2917  */
2918 /*
2919  * Find the most ``active'' lwp of a process and return it for ps display
2920  * purposes
2921  */
2922 static struct lwp *
2923 proc_active_lwp(struct proc *p)
2924 {
2925 	static const int ostat[] = {
2926 		0,
2927 		2,	/* LSIDL */
2928 		6,	/* LSRUN */
2929 		5,	/* LSSLEEP */
2930 		4,	/* LSSTOP */
2931 		0,	/* LSZOMB */
2932 		1,	/* LSDEAD */
2933 		7,	/* LSONPROC */
2934 		3	/* LSSUSPENDED */
2935 	};
2936 
2937 	struct lwp *l, *lp = NULL;
2938 	LIST_FOREACH(l, &p->p_lwps, l_sibling) {
2939 		KASSERT(l->l_stat >= 0 && l->l_stat < __arraycount(ostat));
2940 		if (lp == NULL ||
2941 		    ostat[l->l_stat] > ostat[lp->l_stat] ||
2942 		    (ostat[l->l_stat] == ostat[lp->l_stat] &&
2943 		    l->l_cpticks > lp->l_cpticks)) {
2944 			lp = l;
2945 			continue;
2946 		}
2947 	}
2948 	return lp;
2949 }
2950 
2951 
2952 /*
2953  * Fill in a kinfo_proc2 structure for the specified process.
2954  */
2955 static void
2956 fill_kproc2(struct proc *p, struct kinfo_proc2 *ki, bool zombie)
2957 {
2958 	struct tty *tp;
2959 	struct lwp *l, *l2;
2960 	struct timeval ut, st, rt;
2961 	sigset_t ss1, ss2;
2962 	struct rusage ru;
2963 	struct vmspace *vm;
2964 
2965 	KASSERT(mutex_owned(proc_lock));
2966 	KASSERT(mutex_owned(p->p_lock));
2967 
2968 	sigemptyset(&ss1);
2969 	sigemptyset(&ss2);
2970 	memset(ki, 0, sizeof(*ki));
2971 
2972 	ki->p_paddr = PTRTOUINT64(p);
2973 	ki->p_fd = PTRTOUINT64(p->p_fd);
2974 	ki->p_cwdi = PTRTOUINT64(p->p_cwdi);
2975 	ki->p_stats = PTRTOUINT64(p->p_stats);
2976 	ki->p_limit = PTRTOUINT64(p->p_limit);
2977 	ki->p_vmspace = PTRTOUINT64(p->p_vmspace);
2978 	ki->p_sigacts = PTRTOUINT64(p->p_sigacts);
2979 	ki->p_sess = PTRTOUINT64(p->p_session);
2980 	ki->p_tsess = 0;	/* may be changed if controlling tty below */
2981 	ki->p_ru = PTRTOUINT64(&p->p_stats->p_ru);
2982 	ki->p_eflag = 0;
2983 	ki->p_exitsig = p->p_exitsig;
2984 	ki->p_flag = sysctl_map_flags(sysctl_flagmap, p->p_flag);
2985 	ki->p_flag |= sysctl_map_flags(sysctl_sflagmap, p->p_sflag);
2986 	ki->p_flag |= sysctl_map_flags(sysctl_slflagmap, p->p_slflag);
2987 	ki->p_flag |= sysctl_map_flags(sysctl_lflagmap, p->p_lflag);
2988 	ki->p_flag |= sysctl_map_flags(sysctl_stflagmap, p->p_stflag);
2989 	ki->p_pid = p->p_pid;
2990 	if (p->p_pptr)
2991 		ki->p_ppid = p->p_pptr->p_pid;
2992 	else
2993 		ki->p_ppid = 0;
2994 	ki->p_uid = kauth_cred_geteuid(p->p_cred);
2995 	ki->p_ruid = kauth_cred_getuid(p->p_cred);
2996 	ki->p_gid = kauth_cred_getegid(p->p_cred);
2997 	ki->p_rgid = kauth_cred_getgid(p->p_cred);
2998 	ki->p_svuid = kauth_cred_getsvuid(p->p_cred);
2999 	ki->p_svgid = kauth_cred_getsvgid(p->p_cred);
3000 	ki->p_ngroups = kauth_cred_ngroups(p->p_cred);
3001 	kauth_cred_getgroups(p->p_cred, ki->p_groups,
3002 	    min(ki->p_ngroups, sizeof(ki->p_groups) / sizeof(ki->p_groups[0])),
3003 	    UIO_SYSSPACE);
3004 
3005 	ki->p_uticks = p->p_uticks;
3006 	ki->p_sticks = p->p_sticks;
3007 	ki->p_iticks = p->p_iticks;
3008 	ki->p_tpgid = NO_PGID;	/* may be changed if controlling tty below */
3009 	ki->p_tracep = PTRTOUINT64(p->p_tracep);
3010 	ki->p_traceflag = p->p_traceflag;
3011 
3012 	memcpy(&ki->p_sigignore, &p->p_sigctx.ps_sigignore,sizeof(ki_sigset_t));
3013 	memcpy(&ki->p_sigcatch, &p->p_sigctx.ps_sigcatch, sizeof(ki_sigset_t));
3014 
3015 	ki->p_cpticks = 0;
3016 	ki->p_pctcpu = p->p_pctcpu;
3017 	ki->p_estcpu = 0;
3018 	ki->p_stat = p->p_stat; /* Will likely be overridden by LWP status */
3019 	ki->p_realstat = p->p_stat;
3020 	ki->p_nice = p->p_nice;
3021 	ki->p_xstat = p->p_xstat;
3022 	ki->p_acflag = p->p_acflag;
3023 
3024 	strncpy(ki->p_comm, p->p_comm,
3025 	    min(sizeof(ki->p_comm), sizeof(p->p_comm)));
3026 	strncpy(ki->p_ename, p->p_emul->e_name, sizeof(ki->p_ename));
3027 
3028 	ki->p_nlwps = p->p_nlwps;
3029 	ki->p_realflag = ki->p_flag;
3030 
3031 	if (p->p_stat != SIDL && !P_ZOMBIE(p) && !zombie) {
3032 		vm = p->p_vmspace;
3033 		ki->p_vm_rssize = vm_resident_count(vm);
3034 		ki->p_vm_tsize = vm->vm_tsize;
3035 		ki->p_vm_dsize = vm->vm_dsize;
3036 		ki->p_vm_ssize = vm->vm_ssize;
3037 
3038 		/* Pick the primary (first) LWP */
3039 		l = proc_active_lwp(p);
3040 		KASSERT(l != NULL);
3041 		lwp_lock(l);
3042 		ki->p_nrlwps = p->p_nrlwps;
3043 		ki->p_forw = 0;
3044 		ki->p_back = 0;
3045 		ki->p_addr = PTRTOUINT64(l->l_addr);
3046 		ki->p_stat = l->l_stat;
3047 		ki->p_flag |= sysctl_map_flags(sysctl_lwpflagmap, l->l_flag);
3048 		ki->p_swtime = l->l_swtime;
3049 		ki->p_slptime = l->l_slptime;
3050 		if (l->l_stat == LSONPROC)
3051 			ki->p_schedflags = l->l_cpu->ci_schedstate.spc_flags;
3052 		else
3053 			ki->p_schedflags = 0;
3054 		ki->p_holdcnt = l->l_holdcnt;
3055 		ki->p_priority = lwp_eprio(l);
3056 		ki->p_usrpri = l->l_priority;
3057 		if (l->l_wchan)
3058 			strncpy(ki->p_wmesg, l->l_wmesg, sizeof(ki->p_wmesg));
3059 		ki->p_wchan = PTRTOUINT64(l->l_wchan);
3060 		ki->p_cpuid = cpu_index(l->l_cpu);
3061 		lwp_unlock(l);
3062 		LIST_FOREACH(l, &p->p_lwps, l_sibling) {
3063 			/* This is hardly correct, but... */
3064 			sigplusset(&l->l_sigpend.sp_set, &ss1);
3065 			sigplusset(&l->l_sigmask, &ss2);
3066 			ki->p_cpticks += l->l_cpticks;
3067 			ki->p_pctcpu += l->l_pctcpu;
3068 			ki->p_estcpu += l->l_estcpu;
3069 		}
3070 	}
3071 	sigplusset(&p->p_sigpend.sp_set, &ss2);
3072 	memcpy(&ki->p_siglist, &ss1, sizeof(ki_sigset_t));
3073 	memcpy(&ki->p_sigmask, &ss2, sizeof(ki_sigset_t));
3074 
3075 	if (p->p_session != NULL) {
3076 		ki->p_sid = p->p_session->s_sid;
3077 		ki->p__pgid = p->p_pgrp->pg_id;
3078 		if (p->p_session->s_ttyvp)
3079 			ki->p_eflag |= EPROC_CTTY;
3080 		if (SESS_LEADER(p))
3081 			ki->p_eflag |= EPROC_SLEADER;
3082 		strncpy(ki->p_login, p->p_session->s_login,
3083 		    min(sizeof ki->p_login - 1, sizeof p->p_session->s_login));
3084 		ki->p_jobc = p->p_pgrp->pg_jobc;
3085 		if ((p->p_lflag & PL_CONTROLT) && (tp = p->p_session->s_ttyp)) {
3086 			ki->p_tdev = tp->t_dev;
3087 			ki->p_tpgid = tp->t_pgrp ? tp->t_pgrp->pg_id : NO_PGID;
3088 			ki->p_tsess = PTRTOUINT64(tp->t_session);
3089 		} else {
3090 			ki->p_tdev = (int32_t)NODEV;
3091 		}
3092 	}
3093 
3094 	if (!P_ZOMBIE(p) && !zombie) {
3095 		ki->p_uvalid = 1;
3096 		ki->p_ustart_sec = p->p_stats->p_start.tv_sec;
3097 		ki->p_ustart_usec = p->p_stats->p_start.tv_usec;
3098 
3099 		calcru(p, &ut, &st, NULL, &rt);
3100 		ki->p_rtime_sec = rt.tv_sec;
3101 		ki->p_rtime_usec = rt.tv_usec;
3102 		ki->p_uutime_sec = ut.tv_sec;
3103 		ki->p_uutime_usec = ut.tv_usec;
3104 		ki->p_ustime_sec = st.tv_sec;
3105 		ki->p_ustime_usec = st.tv_usec;
3106 
3107 		memcpy(&ru, &p->p_stats->p_ru, sizeof(ru));
3108 		ki->p_uru_nvcsw = 0;
3109 		ki->p_uru_nivcsw = 0;
3110 		LIST_FOREACH(l2, &p->p_lwps, l_sibling) {
3111 			ki->p_uru_nvcsw += (l2->l_ncsw - l2->l_nivcsw);
3112 			ki->p_uru_nivcsw += l2->l_nivcsw;
3113 			ruadd(&ru, &l2->l_ru);
3114 		}
3115 		ki->p_uru_maxrss = ru.ru_maxrss;
3116 		ki->p_uru_ixrss = ru.ru_ixrss;
3117 		ki->p_uru_idrss = ru.ru_idrss;
3118 		ki->p_uru_isrss = ru.ru_isrss;
3119 		ki->p_uru_minflt = ru.ru_minflt;
3120 		ki->p_uru_majflt = ru.ru_majflt;
3121 		ki->p_uru_nswap = ru.ru_nswap;
3122 		ki->p_uru_inblock = ru.ru_inblock;
3123 		ki->p_uru_oublock = ru.ru_oublock;
3124 		ki->p_uru_msgsnd = ru.ru_msgsnd;
3125 		ki->p_uru_msgrcv = ru.ru_msgrcv;
3126 		ki->p_uru_nsignals = ru.ru_nsignals;
3127 
3128 		timeradd(&p->p_stats->p_cru.ru_utime,
3129 			 &p->p_stats->p_cru.ru_stime, &ut);
3130 		ki->p_uctime_sec = ut.tv_sec;
3131 		ki->p_uctime_usec = ut.tv_usec;
3132 	}
3133 }
3134 
3135 /*
3136  * Fill in a kinfo_lwp structure for the specified lwp.
3137  */
3138 static void
3139 fill_lwp(struct lwp *l, struct kinfo_lwp *kl)
3140 {
3141 	struct proc *p = l->l_proc;
3142 	struct timeval tv;
3143 
3144 	KASSERT(lwp_locked(l, NULL));
3145 
3146 	kl->l_forw = 0;
3147 	kl->l_back = 0;
3148 	kl->l_laddr = PTRTOUINT64(l);
3149 	kl->l_addr = PTRTOUINT64(l->l_addr);
3150 	kl->l_stat = l->l_stat;
3151 	kl->l_lid = l->l_lid;
3152 	kl->l_flag = sysctl_map_flags(sysctl_lwpprflagmap, l->l_prflag);
3153 	kl->l_flag |= sysctl_map_flags(sysctl_lwpflagmap, l->l_flag);
3154 
3155 	kl->l_swtime = l->l_swtime;
3156 	kl->l_slptime = l->l_slptime;
3157 	if (l->l_stat == LSONPROC)
3158 		kl->l_schedflags = l->l_cpu->ci_schedstate.spc_flags;
3159 	else
3160 		kl->l_schedflags = 0;
3161 	kl->l_holdcnt = l->l_holdcnt;
3162 	kl->l_priority = lwp_eprio(l);
3163 	kl->l_usrpri = l->l_priority;
3164 	if (l->l_wchan)
3165 		strncpy(kl->l_wmesg, l->l_wmesg, sizeof(kl->l_wmesg));
3166 	kl->l_wchan = PTRTOUINT64(l->l_wchan);
3167 	kl->l_cpuid = cpu_index(l->l_cpu);
3168 	bintime2timeval(&l->l_rtime, &tv);
3169 	kl->l_rtime_sec = tv.tv_sec;
3170 	kl->l_rtime_usec = tv.tv_usec;
3171 	kl->l_cpticks = l->l_cpticks;
3172 	kl->l_pctcpu = l->l_pctcpu;
3173 	kl->l_pid = p->p_pid;
3174 	if (l->l_name == NULL)
3175 		kl->l_name[0] = '\0';
3176 	else
3177 		strlcpy(kl->l_name, l->l_name, sizeof(kl->l_name));
3178 }
3179 
3180 /*
3181  * Fill in an eproc structure for the specified process.
3182  */
3183 void
3184 fill_eproc(struct proc *p, struct eproc *ep, bool zombie)
3185 {
3186 	struct tty *tp;
3187 	struct lwp *l;
3188 
3189 	KASSERT(mutex_owned(proc_lock));
3190 	KASSERT(mutex_owned(p->p_lock));
3191 
3192 	memset(ep, 0, sizeof(*ep));
3193 
3194 	ep->e_paddr = p;
3195 	ep->e_sess = p->p_session;
3196 	if (p->p_cred) {
3197 		kauth_cred_topcred(p->p_cred, &ep->e_pcred);
3198 		kauth_cred_toucred(p->p_cred, &ep->e_ucred);
3199 	}
3200 	if (p->p_stat != SIDL && !P_ZOMBIE(p) && !zombie) {
3201 		struct vmspace *vm = p->p_vmspace;
3202 
3203 		ep->e_vm.vm_rssize = vm_resident_count(vm);
3204 		ep->e_vm.vm_tsize = vm->vm_tsize;
3205 		ep->e_vm.vm_dsize = vm->vm_dsize;
3206 		ep->e_vm.vm_ssize = vm->vm_ssize;
3207 
3208 		/* Pick the primary (first) LWP */
3209 		l = proc_active_lwp(p);
3210 		KASSERT(l != NULL);
3211 		lwp_lock(l);
3212 		if (l->l_wchan)
3213 			strncpy(ep->e_wmesg, l->l_wmesg, WMESGLEN);
3214 		lwp_unlock(l);
3215 	}
3216 	if (p->p_pptr)
3217 		ep->e_ppid = p->p_pptr->p_pid;
3218 	if (p->p_pgrp && p->p_session) {
3219 		ep->e_pgid = p->p_pgrp->pg_id;
3220 		ep->e_jobc = p->p_pgrp->pg_jobc;
3221 		ep->e_sid = p->p_session->s_sid;
3222 		if ((p->p_lflag & PL_CONTROLT) &&
3223 		    (tp = ep->e_sess->s_ttyp)) {
3224 			ep->e_tdev = tp->t_dev;
3225 			ep->e_tpgid = tp->t_pgrp ? tp->t_pgrp->pg_id : NO_PGID;
3226 			ep->e_tsess = tp->t_session;
3227 		} else
3228 			ep->e_tdev = (uint32_t)NODEV;
3229 		ep->e_flag = ep->e_sess->s_ttyvp ? EPROC_CTTY : 0;
3230 		if (SESS_LEADER(p))
3231 			ep->e_flag |= EPROC_SLEADER;
3232 		strncpy(ep->e_login, ep->e_sess->s_login, MAXLOGNAME);
3233 	}
3234 	ep->e_xsize = ep->e_xrssize = 0;
3235 	ep->e_xccount = ep->e_xswrss = 0;
3236 }
3237 
3238 u_int
3239 sysctl_map_flags(const u_int *map, u_int word)
3240 {
3241 	u_int rv;
3242 
3243 	for (rv = 0; *map != 0; map += 2)
3244 		if ((word & map[0]) != 0)
3245 			rv |= map[1];
3246 
3247 	return rv;
3248 }
3249