xref: /netbsd-src/sys/kern/init_sysctl.c (revision c0179c282a5968435315a82f4128c61372c68fc3)
1 /*	$NetBSD: init_sysctl.c,v 1.91 2006/11/01 22:27:43 christos Exp $ */
2 
3 /*-
4  * Copyright (c) 2003 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.
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  * 3. All advertising materials mentioning features or use of this software
19  *    must display the following acknowledgement:
20  *      This product includes software developed by the NetBSD
21  *      Foundation, Inc. and its contributors.
22  * 4. Neither the name of The NetBSD Foundation nor the names of its
23  *    contributors may be used to endorse or promote products derived
24  *    from this software without specific prior written permission.
25  *
26  * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS
27  * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
28  * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
29  * PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS
30  * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
31  * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
32  * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
33  * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
34  * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
35  * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
36  * POSSIBILITY OF SUCH DAMAGE.
37  */
38 
39 #include <sys/cdefs.h>
40 __KERNEL_RCSID(0, "$NetBSD: init_sysctl.c,v 1.91 2006/11/01 22:27:43 christos Exp $");
41 
42 #include "opt_sysv.h"
43 #include "opt_multiprocessor.h"
44 #include "opt_posix.h"
45 #include "opt_compat_netbsd32.h"
46 #include "opt_ktrace.h"
47 #include "veriexec.h"
48 #include "pty.h"
49 #include "rnd.h"
50 
51 #include <sys/types.h>
52 #include <sys/param.h>
53 #include <sys/sysctl.h>
54 #include <sys/errno.h>
55 #include <sys/systm.h>
56 #include <sys/kernel.h>
57 #include <sys/unistd.h>
58 #include <sys/disklabel.h>
59 #include <sys/rnd.h>
60 #include <sys/vnode.h>
61 #include <sys/mount.h>
62 #include <sys/namei.h>
63 #include <sys/msgbuf.h>
64 #include <dev/cons.h>
65 #include <sys/socketvar.h>
66 #include <sys/file.h>
67 #include <sys/filedesc.h>
68 #include <sys/tty.h>
69 #include <sys/malloc.h>
70 #include <sys/resource.h>
71 #include <sys/resourcevar.h>
72 #include <sys/exec.h>
73 #include <sys/conf.h>
74 #include <sys/device.h>
75 #if NVERIEXEC > 0
76 #define	VERIEXEC_NEED_NODE
77 #include <sys/verified_exec.h>
78 #endif /* NVERIEXEC > 0 */
79 #include <sys/stat.h>
80 #include <sys/kauth.h>
81 #ifdef KTRACE
82 #include <sys/ktrace.h>
83 #endif
84 
85 #if defined(SYSVMSG) || defined(SYSVSEM) || defined(SYSVSHM)
86 #include <sys/ipc.h>
87 #endif
88 #ifdef SYSVMSG
89 #include <sys/msg.h>
90 #endif
91 #ifdef SYSVSEM
92 #include <sys/sem.h>
93 #endif
94 #ifdef SYSVSHM
95 #include <sys/shm.h>
96 #endif
97 
98 #ifdef COMPAT_NETBSD32
99 #include <compat/netbsd32/netbsd32.h>
100 #endif
101 
102 #include <machine/cpu.h>
103 
104 /* XXX this should not be here */
105 int security_setidcore_dump;
106 char security_setidcore_path[MAXPATHLEN] = "/var/crash/%n.core";
107 uid_t security_setidcore_owner = 0;
108 gid_t security_setidcore_group = 0;
109 mode_t security_setidcore_mode = (S_IRUSR|S_IWUSR);
110 
111 /*
112  * try over estimating by 5 procs/lwps
113  */
114 #define KERN_PROCSLOP	(5 * sizeof(struct kinfo_proc))
115 #define KERN_LWPSLOP	(5 * sizeof(struct kinfo_lwp))
116 
117 #ifdef KTRACE
118 int dcopyout(struct lwp *, const void *, void *, size_t);
119 
120 int
121 dcopyout(l, kaddr, uaddr, len)
122 	struct lwp *l;
123 	const void *kaddr;
124 	void *uaddr;
125 	size_t len;
126 {
127 	int error;
128 
129 	error = copyout(kaddr, uaddr, len);
130 	if (!error && KTRPOINT(l->l_proc, KTR_MIB)) {
131 		struct iovec iov;
132 
133 		iov.iov_base = uaddr;
134 		iov.iov_len = len;
135 		ktrgenio(l, -1, UIO_READ, &iov, len, 0);
136 	}
137 	return error;
138 }
139 #else /* !KTRACE */
140 #define dcopyout(l, kaddr, uaddr, len) copyout(kaddr, uaddr, len)
141 #endif /* KTRACE */
142 #ifndef MULTIPROCESSOR
143 #define	sysctl_ncpus()	(1)
144 #else /* MULTIPROCESSOR */
145 #ifndef CPU_INFO_FOREACH
146 #define CPU_INFO_ITERATOR int
147 #define CPU_INFO_FOREACH(cii, ci) cii = 0, ci = curcpu(); ci != NULL; ci = NULL
148 #endif
149 static int
150 sysctl_ncpus(void)
151 {
152 	struct cpu_info *ci;
153 	CPU_INFO_ITERATOR cii;
154 
155 	int ncpus = 0;
156 	for (CPU_INFO_FOREACH(cii, ci))
157 		ncpus++;
158 	return (ncpus);
159 }
160 #endif /* MULTIPROCESSOR */
161 
162 #ifdef DIAGNOSTIC
163 static int sysctl_kern_trigger_panic(SYSCTLFN_PROTO);
164 #endif
165 static int sysctl_kern_maxvnodes(SYSCTLFN_PROTO);
166 static int sysctl_kern_rtc_offset(SYSCTLFN_PROTO);
167 static int sysctl_kern_maxproc(SYSCTLFN_PROTO);
168 static int sysctl_kern_hostid(SYSCTLFN_PROTO);
169 static int sysctl_setlen(SYSCTLFN_PROTO);
170 static int sysctl_kern_clockrate(SYSCTLFN_PROTO);
171 static int sysctl_kern_file(SYSCTLFN_PROTO);
172 static int sysctl_kern_autonice(SYSCTLFN_PROTO);
173 static int sysctl_msgbuf(SYSCTLFN_PROTO);
174 static int sysctl_kern_defcorename(SYSCTLFN_PROTO);
175 static int sysctl_kern_cptime(SYSCTLFN_PROTO);
176 #if defined(SYSVMSG) || defined(SYSVSEM) || defined(SYSVSHM)
177 static int sysctl_kern_sysvipc(SYSCTLFN_PROTO);
178 #endif /* defined(SYSVMSG) || defined(SYSVSEM) || defined(SYSVSHM) */
179 #if NPTY > 0
180 static int sysctl_kern_maxptys(SYSCTLFN_PROTO);
181 #endif /* NPTY > 0 */
182 static int sysctl_kern_sbmax(SYSCTLFN_PROTO);
183 static int sysctl_kern_urnd(SYSCTLFN_PROTO);
184 static int sysctl_kern_arnd(SYSCTLFN_PROTO);
185 static int sysctl_kern_lwp(SYSCTLFN_PROTO);
186 static int sysctl_kern_forkfsleep(SYSCTLFN_PROTO);
187 static int sysctl_kern_root_partition(SYSCTLFN_PROTO);
188 static int sysctl_kern_drivers(SYSCTLFN_PROTO);
189 static int sysctl_kern_file2(SYSCTLFN_PROTO);
190 #if NVERIEXEC > 0
191 static int sysctl_kern_veriexec(SYSCTLFN_PROTO);
192 #endif /* NVERIEXEC > 0 */
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 static int sysctl_hw_ncpu(SYSCTLFN_PROTO);
201 
202 static void fill_kproc2(struct proc *, struct kinfo_proc2 *);
203 static void fill_lwp(struct lwp *l, struct kinfo_lwp *kl);
204 static void fill_file(struct kinfo_file *, const struct file *, struct proc *,
205 		      int);
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 	sysctl_createv(clog, 0, NULL, NULL,
454 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
455 		       CTLTYPE_STRING, "domainname",
456 		       SYSCTL_DESCR("YP domain name"),
457 		       sysctl_setlen, 0, &domainname, MAXHOSTNAMELEN,
458 		       CTL_KERN, KERN_DOMAINNAME, CTL_EOL);
459 	sysctl_createv(clog, 0, NULL, NULL,
460 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
461 		       CTLTYPE_INT, "maxpartitions",
462 		       SYSCTL_DESCR("Maximum number of partitions allowed per "
463 				    "disk"),
464 		       NULL, MAXPARTITIONS, NULL, 0,
465 		       CTL_KERN, KERN_MAXPARTITIONS, CTL_EOL);
466 	sysctl_createv(clog, 0, NULL, NULL,
467 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
468 		       CTLTYPE_INT, "rawpartition",
469 		       SYSCTL_DESCR("Raw partition of a disk"),
470 		       NULL, RAW_PART, NULL, 0,
471 		       CTL_KERN, KERN_RAWPARTITION, CTL_EOL);
472 	sysctl_createv(clog, 0, NULL, NULL,
473 		       CTLFLAG_PERMANENT,
474 		       CTLTYPE_STRUCT, "timex", NULL,
475 		       sysctl_notavail, 0, NULL, 0,
476 		       CTL_KERN, KERN_TIMEX, CTL_EOL);
477 	sysctl_createv(clog, 0, NULL, NULL,
478 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
479 		       CTLTYPE_INT, "autonicetime",
480 		       SYSCTL_DESCR("CPU clock seconds before non-root "
481 				    "process priority is lowered"),
482 		       sysctl_kern_autonice, 0, &autonicetime, 0,
483 		       CTL_KERN, KERN_AUTONICETIME, CTL_EOL);
484 	sysctl_createv(clog, 0, NULL, NULL,
485 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
486 		       CTLTYPE_INT, "autoniceval",
487 		       SYSCTL_DESCR("Automatic reniced non-root process "
488 				    "priority"),
489 		       sysctl_kern_autonice, 0, &autoniceval, 0,
490 		       CTL_KERN, KERN_AUTONICEVAL, CTL_EOL);
491 	sysctl_createv(clog, 0, NULL, NULL,
492 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
493 		       CTLTYPE_INT, "rtc_offset",
494 		       SYSCTL_DESCR("Offset of real time clock from UTC in "
495 				    "minutes"),
496 		       sysctl_kern_rtc_offset, 0, &rtc_offset, 0,
497 		       CTL_KERN, KERN_RTC_OFFSET, CTL_EOL);
498 	sysctl_createv(clog, 0, NULL, NULL,
499 		       CTLFLAG_PERMANENT,
500 		       CTLTYPE_STRING, "root_device",
501 		       SYSCTL_DESCR("Name of the root device"),
502 		       sysctl_root_device, 0, NULL, 0,
503 		       CTL_KERN, KERN_ROOT_DEVICE, CTL_EOL);
504 	sysctl_createv(clog, 0, NULL, NULL,
505 		       CTLFLAG_PERMANENT,
506 		       CTLTYPE_INT, "msgbufsize",
507 		       SYSCTL_DESCR("Size of the kernel message buffer"),
508 		       sysctl_msgbuf, 0, NULL, 0,
509 		       CTL_KERN, KERN_MSGBUFSIZE, CTL_EOL);
510 	sysctl_createv(clog, 0, NULL, NULL,
511 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
512 		       CTLTYPE_INT, "fsync",
513 		       SYSCTL_DESCR("Whether the POSIX 1003.1b File "
514 				    "Synchronization Option is available on "
515 				    "this system"),
516 		       NULL, 1, NULL, 0,
517 		       CTL_KERN, KERN_FSYNC, CTL_EOL);
518 	sysctl_createv(clog, 0, NULL, NULL,
519 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
520 		       CTLTYPE_INT, "sysvmsg",
521 		       SYSCTL_DESCR("System V style message support available"),
522 		       NULL,
523 #ifdef SYSVMSG
524 		       1,
525 #else /* SYSVMSG */
526 		       0,
527 #endif /* SYSVMSG */
528 		       NULL, 0, CTL_KERN, KERN_SYSVMSG, CTL_EOL);
529 	sysctl_createv(clog, 0, NULL, NULL,
530 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
531 		       CTLTYPE_INT, "sysvsem",
532 		       SYSCTL_DESCR("System V style semaphore support "
533 				    "available"), NULL,
534 #ifdef SYSVSEM
535 		       1,
536 #else /* SYSVSEM */
537 		       0,
538 #endif /* SYSVSEM */
539 		       NULL, 0, CTL_KERN, KERN_SYSVSEM, CTL_EOL);
540 	sysctl_createv(clog, 0, NULL, NULL,
541 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
542 		       CTLTYPE_INT, "sysvshm",
543 		       SYSCTL_DESCR("System V style shared memory support "
544 				    "available"), NULL,
545 #ifdef SYSVSHM
546 		       1,
547 #else /* SYSVSHM */
548 		       0,
549 #endif /* SYSVSHM */
550 		       NULL, 0, CTL_KERN, KERN_SYSVSHM, CTL_EOL);
551 	sysctl_createv(clog, 0, NULL, NULL,
552 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
553 		       CTLTYPE_INT, "synchronized_io",
554 		       SYSCTL_DESCR("Whether the POSIX 1003.1b Synchronized "
555 				    "I/O Option is available on this system"),
556 		       NULL, 1, NULL, 0,
557 		       CTL_KERN, KERN_SYNCHRONIZED_IO, CTL_EOL);
558 	sysctl_createv(clog, 0, NULL, NULL,
559 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
560 		       CTLTYPE_INT, "iov_max",
561 		       SYSCTL_DESCR("Maximum number of iovec structures per "
562 				    "process"),
563 		       NULL, IOV_MAX, NULL, 0,
564 		       CTL_KERN, KERN_IOV_MAX, CTL_EOL);
565 	sysctl_createv(clog, 0, NULL, NULL,
566 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
567 		       CTLTYPE_INT, "mapped_files",
568 		       SYSCTL_DESCR("Whether the POSIX 1003.1b Memory Mapped "
569 				    "Files Option is available on this system"),
570 		       NULL, 1, NULL, 0,
571 		       CTL_KERN, KERN_MAPPED_FILES, CTL_EOL);
572 	sysctl_createv(clog, 0, NULL, NULL,
573 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
574 		       CTLTYPE_INT, "memlock",
575 		       SYSCTL_DESCR("Whether the POSIX 1003.1b Process Memory "
576 				    "Locking Option is available on this "
577 				    "system"),
578 		       NULL, 1, NULL, 0,
579 		       CTL_KERN, KERN_MEMLOCK, CTL_EOL);
580 	sysctl_createv(clog, 0, NULL, NULL,
581 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
582 		       CTLTYPE_INT, "memlock_range",
583 		       SYSCTL_DESCR("Whether the POSIX 1003.1b Range Memory "
584 				    "Locking Option is available on this "
585 				    "system"),
586 		       NULL, 1, NULL, 0,
587 		       CTL_KERN, KERN_MEMLOCK_RANGE, CTL_EOL);
588 	sysctl_createv(clog, 0, NULL, NULL,
589 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
590 		       CTLTYPE_INT, "memory_protection",
591 		       SYSCTL_DESCR("Whether the POSIX 1003.1b Memory "
592 				    "Protection Option is available on this "
593 				    "system"),
594 		       NULL, 1, NULL, 0,
595 		       CTL_KERN, KERN_MEMORY_PROTECTION, CTL_EOL);
596 	sysctl_createv(clog, 0, NULL, NULL,
597 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
598 		       CTLTYPE_INT, "login_name_max",
599 		       SYSCTL_DESCR("Maximum login name length"),
600 		       NULL, LOGIN_NAME_MAX, NULL, 0,
601 		       CTL_KERN, KERN_LOGIN_NAME_MAX, CTL_EOL);
602 	sysctl_createv(clog, 0, NULL, NULL,
603 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
604 		       CTLTYPE_STRING, "defcorename",
605 		       SYSCTL_DESCR("Default core file name"),
606 		       sysctl_kern_defcorename, 0, defcorename, MAXPATHLEN,
607 		       CTL_KERN, KERN_DEFCORENAME, CTL_EOL);
608 	sysctl_createv(clog, 0, NULL, NULL,
609 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
610 		       CTLTYPE_INT, "logsigexit",
611 		       SYSCTL_DESCR("Log process exit when caused by signals"),
612 		       NULL, 0, &kern_logsigexit, 0,
613 		       CTL_KERN, KERN_LOGSIGEXIT, CTL_EOL);
614 	sysctl_createv(clog, 0, NULL, NULL,
615 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
616 		       CTLTYPE_INT, "fscale",
617 		       SYSCTL_DESCR("Kernel fixed-point scale factor"),
618 		       NULL, FSCALE, NULL, 0,
619 		       CTL_KERN, KERN_FSCALE, CTL_EOL);
620 	sysctl_createv(clog, 0, NULL, NULL,
621 		       CTLFLAG_PERMANENT,
622 		       CTLTYPE_INT, "ccpu",
623 		       SYSCTL_DESCR("Scheduler exponential decay value"),
624 		       NULL, 0, &ccpu, 0,
625 		       CTL_KERN, KERN_CCPU, CTL_EOL);
626 	sysctl_createv(clog, 0, NULL, NULL,
627 		       CTLFLAG_PERMANENT,
628 		       CTLTYPE_STRUCT, "cp_time",
629 		       SYSCTL_DESCR("Clock ticks spent in different CPU states"),
630 		       sysctl_kern_cptime, 0, NULL, 0,
631 		       CTL_KERN, KERN_CP_TIME, CTL_EOL);
632 #if defined(SYSVMSG) || defined(SYSVSEM) || defined(SYSVSHM)
633 	sysctl_createv(clog, 0, NULL, NULL,
634 		       CTLFLAG_PERMANENT,
635 		       CTLTYPE_STRUCT, "sysvipc_info",
636 		       SYSCTL_DESCR("System V style IPC information"),
637 		       sysctl_kern_sysvipc, 0, NULL, 0,
638 		       CTL_KERN, KERN_SYSVIPC_INFO, CTL_EOL);
639 #endif /* SYSVMSG || SYSVSEM || SYSVSHM */
640 	sysctl_createv(clog, 0, NULL, NULL,
641 		       CTLFLAG_PERMANENT,
642 		       CTLTYPE_INT, "msgbuf",
643 		       SYSCTL_DESCR("Kernel message buffer"),
644 		       sysctl_msgbuf, 0, NULL, 0,
645 		       CTL_KERN, KERN_MSGBUF, CTL_EOL);
646 	sysctl_createv(clog, 0, NULL, NULL,
647 		       CTLFLAG_PERMANENT,
648 		       CTLTYPE_STRUCT, "consdev",
649 		       SYSCTL_DESCR("Console device"),
650 		       sysctl_consdev, 0, NULL, sizeof(dev_t),
651 		       CTL_KERN, KERN_CONSDEV, CTL_EOL);
652 #if NPTY > 0
653 	sysctl_createv(clog, 0, NULL, NULL,
654 		       CTLFLAG_PERMANENT,
655 		       CTLTYPE_INT, "maxptys",
656 		       SYSCTL_DESCR("Maximum number of pseudo-ttys"),
657 		       sysctl_kern_maxptys, 0, NULL, 0,
658 		       CTL_KERN, KERN_MAXPTYS, CTL_EOL);
659 #endif /* NPTY > 0 */
660 	sysctl_createv(clog, 0, NULL, NULL,
661 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
662 		       CTLTYPE_INT, "maxphys",
663 		       SYSCTL_DESCR("Maximum raw I/O transfer size"),
664 		       NULL, MAXPHYS, NULL, 0,
665 		       CTL_KERN, KERN_MAXPHYS, CTL_EOL);
666 	sysctl_createv(clog, 0, NULL, NULL,
667 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
668 		       CTLTYPE_INT, "sbmax",
669 		       SYSCTL_DESCR("Maximum socket buffer size"),
670 		       sysctl_kern_sbmax, 0, NULL, 0,
671 		       CTL_KERN, KERN_SBMAX, CTL_EOL);
672 	sysctl_createv(clog, 0, NULL, NULL,
673 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
674 		       CTLTYPE_INT, "monotonic_clock",
675 		       SYSCTL_DESCR("Implementation version of the POSIX "
676 				    "1003.1b Monotonic Clock Option"),
677 		       /* XXX _POSIX_VERSION */
678 		       NULL, _POSIX_MONOTONIC_CLOCK, NULL, 0,
679 		       CTL_KERN, KERN_MONOTONIC_CLOCK, CTL_EOL);
680 	sysctl_createv(clog, 0, NULL, NULL,
681 		       CTLFLAG_PERMANENT,
682 		       CTLTYPE_INT, "urandom",
683 		       SYSCTL_DESCR("Random integer value"),
684 		       sysctl_kern_urnd, 0, NULL, 0,
685 		       CTL_KERN, KERN_URND, CTL_EOL);
686 	sysctl_createv(clog, 0, NULL, NULL,
687 		       CTLFLAG_PERMANENT,
688 		       CTLTYPE_INT, "arandom",
689 		       SYSCTL_DESCR("n bytes of random data"),
690 		       sysctl_kern_arnd, 0, NULL, 0,
691 		       CTL_KERN, KERN_ARND, CTL_EOL);
692 	sysctl_createv(clog, 0, NULL, NULL,
693 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
694 		       CTLTYPE_INT, "labelsector",
695 		       SYSCTL_DESCR("Sector number containing the disklabel"),
696 		       NULL, LABELSECTOR, NULL, 0,
697 		       CTL_KERN, KERN_LABELSECTOR, CTL_EOL);
698 	sysctl_createv(clog, 0, NULL, NULL,
699 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
700 		       CTLTYPE_INT, "labeloffset",
701 		       SYSCTL_DESCR("Offset of the disklabel within the "
702 				    "sector"),
703 		       NULL, LABELOFFSET, NULL, 0,
704 		       CTL_KERN, KERN_LABELOFFSET, CTL_EOL);
705 	sysctl_createv(clog, 0, NULL, NULL,
706 		       CTLFLAG_PERMANENT,
707 		       CTLTYPE_NODE, "lwp",
708 		       SYSCTL_DESCR("System-wide LWP information"),
709 		       sysctl_kern_lwp, 0, NULL, 0,
710 		       CTL_KERN, KERN_LWP, CTL_EOL);
711 	sysctl_createv(clog, 0, NULL, NULL,
712 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
713 		       CTLTYPE_INT, "forkfsleep",
714 		       SYSCTL_DESCR("Milliseconds to sleep on fork failure due "
715 				    "to process limits"),
716 		       sysctl_kern_forkfsleep, 0, NULL, 0,
717 		       CTL_KERN, KERN_FORKFSLEEP, CTL_EOL);
718 	sysctl_createv(clog, 0, NULL, NULL,
719 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
720 		       CTLTYPE_INT, "posix_threads",
721 		       SYSCTL_DESCR("Version of IEEE Std 1003.1 and its "
722 				    "Threads option to which the system "
723 				    "attempts to conform"),
724 		       /* XXX _POSIX_VERSION */
725 		       NULL, _POSIX_THREADS, NULL, 0,
726 		       CTL_KERN, KERN_POSIX_THREADS, CTL_EOL);
727 	sysctl_createv(clog, 0, NULL, NULL,
728 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
729 		       CTLTYPE_INT, "posix_semaphores",
730 		       SYSCTL_DESCR("Version of IEEE Std 1003.1 and its "
731 				    "Semaphores option to which the system "
732 				    "attempts to conform"), NULL,
733 #ifdef P1003_1B_SEMAPHORE
734 		       200112,
735 #else /* P1003_1B_SEMAPHORE */
736 		       0,
737 #endif /* P1003_1B_SEMAPHORE */
738 		       NULL, 0, CTL_KERN, KERN_POSIX_SEMAPHORES, CTL_EOL);
739 	sysctl_createv(clog, 0, NULL, NULL,
740 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
741 		       CTLTYPE_INT, "posix_barriers",
742 		       SYSCTL_DESCR("Version of IEEE Std 1003.1 and its "
743 				    "Barriers option to which the system "
744 				    "attempts to conform"),
745 		       /* XXX _POSIX_VERSION */
746 		       NULL, _POSIX_BARRIERS, NULL, 0,
747 		       CTL_KERN, KERN_POSIX_BARRIERS, CTL_EOL);
748 	sysctl_createv(clog, 0, NULL, NULL,
749 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
750 		       CTLTYPE_INT, "posix_timers",
751 		       SYSCTL_DESCR("Version of IEEE Std 1003.1 and its "
752 				    "Timers option to which the system "
753 				    "attempts to conform"),
754 		       /* XXX _POSIX_VERSION */
755 		       NULL, _POSIX_TIMERS, NULL, 0,
756 		       CTL_KERN, KERN_POSIX_TIMERS, CTL_EOL);
757 	sysctl_createv(clog, 0, NULL, NULL,
758 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
759 		       CTLTYPE_INT, "posix_spin_locks",
760 		       SYSCTL_DESCR("Version of IEEE Std 1003.1 and its Spin "
761 				    "Locks option to which the system attempts "
762 				    "to conform"),
763 		       /* XXX _POSIX_VERSION */
764 		       NULL, _POSIX_SPIN_LOCKS, NULL, 0,
765 		       CTL_KERN, KERN_POSIX_SPIN_LOCKS, CTL_EOL);
766 	sysctl_createv(clog, 0, NULL, NULL,
767 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
768 		       CTLTYPE_INT, "posix_reader_writer_locks",
769 		       SYSCTL_DESCR("Version of IEEE Std 1003.1 and its "
770 				    "Read-Write Locks option to which the "
771 				    "system attempts to conform"),
772 		       /* XXX _POSIX_VERSION */
773 		       NULL, _POSIX_READER_WRITER_LOCKS, NULL, 0,
774 		       CTL_KERN, KERN_POSIX_READER_WRITER_LOCKS, CTL_EOL);
775 	sysctl_createv(clog, 0, NULL, NULL,
776 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
777 		       CTLTYPE_INT, "dump_on_panic",
778 		       SYSCTL_DESCR("Perform a crash dump on system panic"),
779 		       NULL, 0, &dumponpanic, 0,
780 		       CTL_KERN, KERN_DUMP_ON_PANIC, CTL_EOL);
781 #ifdef DIAGNOSTIC
782 	sysctl_createv(clog, 0, NULL, NULL,
783 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
784 		       CTLTYPE_INT, "panic_now",
785 		       SYSCTL_DESCR("Trigger a panic"),
786 		       sysctl_kern_trigger_panic, 0, NULL, 0,
787 		       CTL_KERN, CTL_CREATE, CTL_EOL);
788 #endif
789 	sysctl_createv(clog, 0, NULL, NULL,
790 		       CTLFLAG_PERMANENT,
791 		       CTLTYPE_INT, "root_partition",
792 		       SYSCTL_DESCR("Root partition on the root device"),
793 		       sysctl_kern_root_partition, 0, NULL, 0,
794 		       CTL_KERN, KERN_ROOT_PARTITION, CTL_EOL);
795 	sysctl_createv(clog, 0, NULL, NULL,
796 		       CTLFLAG_PERMANENT,
797 		       CTLTYPE_STRUCT, "drivers",
798 		       SYSCTL_DESCR("List of all drivers with block and "
799 				    "character device numbers"),
800 		       sysctl_kern_drivers, 0, NULL, 0,
801 		       CTL_KERN, KERN_DRIVERS, CTL_EOL);
802 	sysctl_createv(clog, 0, NULL, NULL,
803 		       CTLFLAG_PERMANENT,
804 		       CTLTYPE_STRUCT, "file2",
805 		       SYSCTL_DESCR("System open file table"),
806 		       sysctl_kern_file2, 0, NULL, 0,
807 		       CTL_KERN, KERN_FILE2, CTL_EOL);
808 #if NVERIEXEC > 0
809 	sysctl_createv(clog, 0, NULL, NULL,
810 		       CTLFLAG_PERMANENT,
811 		       CTLTYPE_NODE, "veriexec",
812 		       SYSCTL_DESCR("Verified Exec"),
813 		       NULL, 0, NULL, 0,
814 		       CTL_KERN, KERN_VERIEXEC, CTL_EOL);
815 	sysctl_createv(clog, 0, NULL, NULL,
816 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
817 		       CTLTYPE_INT, "verbose",
818 		       SYSCTL_DESCR("Verified Exec verbose level"),
819 		       NULL, 0, &veriexec_verbose, 0,
820 		       CTL_KERN, KERN_VERIEXEC, VERIEXEC_VERBOSE,
821 		       CTL_EOL);
822 	sysctl_createv(clog, 0, NULL, NULL,
823 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
824 		       CTLTYPE_INT, "strict",
825 		       SYSCTL_DESCR("Verified Exec strict level"),
826 		       sysctl_kern_veriexec, 0, NULL, 0,
827 		       CTL_KERN, KERN_VERIEXEC, VERIEXEC_STRICT, CTL_EOL);
828 	sysctl_createv(clog, 0, NULL, NULL,
829 		       CTLFLAG_PERMANENT,
830 		       CTLTYPE_STRING, "algorithms",
831 		       SYSCTL_DESCR("Verified Exec supported hashing "
832 				    "algorithms"),
833 		       sysctl_kern_veriexec, 0, NULL, 0,
834 		       CTL_KERN, KERN_VERIEXEC, VERIEXEC_ALGORITHMS, CTL_EOL);
835 	sysctl_createv(clog, 0, NULL, &veriexec_count_node,
836 		       CTLFLAG_PERMANENT,
837 		       CTLTYPE_NODE, "count",
838 		       SYSCTL_DESCR("Number of fingerprints on device(s)"),
839 		       NULL, 0, NULL, 0,
840 		       CTL_KERN, KERN_VERIEXEC, VERIEXEC_COUNT, CTL_EOL);
841 #endif /* NVERIEXEC > 0 */
842 	sysctl_createv(clog, 0, NULL, NULL,
843 		       CTLFLAG_PERMANENT,
844 		       CTLTYPE_STRUCT, "cp_id",
845 		       SYSCTL_DESCR("Mapping of CPU number to CPU id"),
846 		       sysctl_kern_cpid, 0, NULL, 0,
847 		       CTL_KERN, KERN_CP_ID, CTL_EOL);
848 
849 	sysctl_createv(clog, 0, NULL, &rnode,
850 		       CTLFLAG_PERMANENT,
851 		       CTLTYPE_NODE, "coredump",
852 		       SYSCTL_DESCR("Coredump settings."),
853 		       NULL, 0, NULL, 0,
854 		       CTL_KERN, CTL_CREATE, CTL_EOL);
855 	sysctl_createv(clog, 0, &rnode, &rnode,
856 		       CTLFLAG_PERMANENT,
857 		       CTLTYPE_NODE, "setid",
858 		       SYSCTL_DESCR("Set-id processes' coredump settings."),
859 		       NULL, 0, NULL, 0,
860 		       CTL_CREATE, CTL_EOL);
861 	sysctl_createv(clog, 0, &rnode, NULL,
862 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
863 		       CTLTYPE_INT, "dump",
864 		       SYSCTL_DESCR("Allow set-id processes to dump core."),
865 		       sysctl_security_setidcore, 0, &security_setidcore_dump,
866 		       sizeof(security_setidcore_dump),
867 		       CTL_CREATE, CTL_EOL);
868 	sysctl_createv(clog, 0, &rnode, NULL,
869 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
870 		       CTLTYPE_STRING, "path",
871 		       SYSCTL_DESCR("Path pattern for set-id coredumps."),
872 		       sysctl_security_setidcorename, 0,
873 		       &security_setidcore_path,
874 		       sizeof(security_setidcore_path),
875 		       CTL_CREATE, CTL_EOL);
876 	sysctl_createv(clog, 0, &rnode, NULL,
877 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
878 		       CTLTYPE_INT, "owner",
879 		       SYSCTL_DESCR("Owner id for set-id processes' cores."),
880 		       sysctl_security_setidcore, 0, &security_setidcore_owner,
881 		       0,
882 		       CTL_CREATE, CTL_EOL);
883 	sysctl_createv(clog, 0, &rnode, NULL,
884 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
885 		       CTLTYPE_INT, "group",
886 		       SYSCTL_DESCR("Group id for set-id processes' cores."),
887 		       sysctl_security_setidcore, 0, &security_setidcore_group,
888 		       0,
889 		       CTL_CREATE, CTL_EOL);
890 	sysctl_createv(clog, 0, &rnode, NULL,
891 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
892 		       CTLTYPE_INT, "mode",
893 		       SYSCTL_DESCR("Mode for set-id processes' cores."),
894 		       sysctl_security_setidcore, 0, &security_setidcore_mode,
895 		       0,
896 		       CTL_CREATE, CTL_EOL);
897 }
898 
899 SYSCTL_SETUP(sysctl_kern_proc_setup,
900 	     "sysctl kern.proc/proc2/proc_args subtree setup")
901 {
902 
903 	sysctl_createv(clog, 0, NULL, NULL,
904 		       CTLFLAG_PERMANENT,
905 		       CTLTYPE_NODE, "kern", NULL,
906 		       NULL, 0, NULL, 0,
907 		       CTL_KERN, CTL_EOL);
908 
909 	sysctl_createv(clog, 0, NULL, NULL,
910 		       CTLFLAG_PERMANENT,
911 		       CTLTYPE_NODE, "proc",
912 		       SYSCTL_DESCR("System-wide process information"),
913 		       sysctl_doeproc, 0, NULL, 0,
914 		       CTL_KERN, KERN_PROC, CTL_EOL);
915 	sysctl_createv(clog, 0, NULL, NULL,
916 		       CTLFLAG_PERMANENT,
917 		       CTLTYPE_NODE, "proc2",
918 		       SYSCTL_DESCR("Machine-independent process information"),
919 		       sysctl_doeproc, 0, NULL, 0,
920 		       CTL_KERN, KERN_PROC2, CTL_EOL);
921 	sysctl_createv(clog, 0, NULL, NULL,
922 		       CTLFLAG_PERMANENT,
923 		       CTLTYPE_NODE, "proc_args",
924 		       SYSCTL_DESCR("Process argument information"),
925 		       sysctl_kern_proc_args, 0, NULL, 0,
926 		       CTL_KERN, KERN_PROC_ARGS, CTL_EOL);
927 
928 	/*
929 	  "nodes" under these:
930 
931 	  KERN_PROC_ALL
932 	  KERN_PROC_PID pid
933 	  KERN_PROC_PGRP pgrp
934 	  KERN_PROC_SESSION sess
935 	  KERN_PROC_TTY tty
936 	  KERN_PROC_UID uid
937 	  KERN_PROC_RUID uid
938 	  KERN_PROC_GID gid
939 	  KERN_PROC_RGID gid
940 
941 	  all in all, probably not worth the effort...
942 	*/
943 }
944 
945 SYSCTL_SETUP(sysctl_hw_setup, "sysctl hw subtree setup")
946 {
947 	u_int u;
948 	u_quad_t q;
949 
950 	sysctl_createv(clog, 0, NULL, NULL,
951 		       CTLFLAG_PERMANENT,
952 		       CTLTYPE_NODE, "hw", NULL,
953 		       NULL, 0, NULL, 0,
954 		       CTL_HW, CTL_EOL);
955 
956 	sysctl_createv(clog, 0, NULL, NULL,
957 		       CTLFLAG_PERMANENT,
958 		       CTLTYPE_STRING, "machine",
959 		       SYSCTL_DESCR("Machine class"),
960 		       NULL, 0, machine, 0,
961 		       CTL_HW, HW_MACHINE, CTL_EOL);
962 	sysctl_createv(clog, 0, NULL, NULL,
963 		       CTLFLAG_PERMANENT,
964 		       CTLTYPE_STRING, "model",
965 		       SYSCTL_DESCR("Machine model"),
966 		       NULL, 0, cpu_model, 0,
967 		       CTL_HW, HW_MODEL, CTL_EOL);
968 	sysctl_createv(clog, 0, NULL, NULL,
969 		       CTLFLAG_PERMANENT,
970 		       CTLTYPE_INT, "ncpu",
971 		       SYSCTL_DESCR("Number of active CPUs"),
972 		       sysctl_hw_ncpu, 0, NULL, 0,
973 		       CTL_HW, HW_NCPU, CTL_EOL);
974 	sysctl_createv(clog, 0, NULL, NULL,
975 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
976 		       CTLTYPE_INT, "byteorder",
977 		       SYSCTL_DESCR("System byte order"),
978 		       NULL, BYTE_ORDER, NULL, 0,
979 		       CTL_HW, HW_BYTEORDER, CTL_EOL);
980 	u = ((u_int)physmem > (UINT_MAX / PAGE_SIZE)) ?
981 		UINT_MAX : physmem * PAGE_SIZE;
982 	sysctl_createv(clog, 0, NULL, NULL,
983 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
984 		       CTLTYPE_INT, "physmem",
985 		       SYSCTL_DESCR("Bytes of physical memory"),
986 		       NULL, u, NULL, 0,
987 		       CTL_HW, HW_PHYSMEM, CTL_EOL);
988 	sysctl_createv(clog, 0, NULL, NULL,
989 		       CTLFLAG_PERMANENT,
990 		       CTLTYPE_INT, "usermem",
991 		       SYSCTL_DESCR("Bytes of non-kernel memory"),
992 		       sysctl_hw_usermem, 0, NULL, 0,
993 		       CTL_HW, HW_USERMEM, CTL_EOL);
994 	sysctl_createv(clog, 0, NULL, NULL,
995 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
996 		       CTLTYPE_INT, "pagesize",
997 		       SYSCTL_DESCR("Software page size"),
998 		       NULL, PAGE_SIZE, NULL, 0,
999 		       CTL_HW, HW_PAGESIZE, CTL_EOL);
1000 	sysctl_createv(clog, 0, NULL, NULL,
1001 		       CTLFLAG_PERMANENT,
1002 		       CTLTYPE_STRING, "machine_arch",
1003 		       SYSCTL_DESCR("Machine CPU class"),
1004 		       NULL, 0, machine_arch, 0,
1005 		       CTL_HW, HW_MACHINE_ARCH, CTL_EOL);
1006 	sysctl_createv(clog, 0, NULL, NULL,
1007 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
1008 		       CTLTYPE_INT, "alignbytes",
1009 		       SYSCTL_DESCR("Alignment constraint for all possible "
1010 				    "data types"),
1011 		       NULL, ALIGNBYTES, NULL, 0,
1012 		       CTL_HW, HW_ALIGNBYTES, CTL_EOL);
1013 	sysctl_createv(clog, 0, NULL, NULL,
1014 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE|CTLFLAG_HEX,
1015 		       CTLTYPE_STRING, "cnmagic",
1016 		       SYSCTL_DESCR("Console magic key sequence"),
1017 		       sysctl_hw_cnmagic, 0, NULL, CNS_LEN,
1018 		       CTL_HW, HW_CNMAGIC, CTL_EOL);
1019 	q = (u_quad_t)physmem * PAGE_SIZE;
1020 	sysctl_createv(clog, 0, NULL, NULL,
1021 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
1022 		       CTLTYPE_QUAD, "physmem64",
1023 		       SYSCTL_DESCR("Bytes of physical memory"),
1024 		       NULL, q, NULL, 0,
1025 		       CTL_HW, HW_PHYSMEM64, CTL_EOL);
1026 	sysctl_createv(clog, 0, NULL, NULL,
1027 		       CTLFLAG_PERMANENT,
1028 		       CTLTYPE_QUAD, "usermem64",
1029 		       SYSCTL_DESCR("Bytes of non-kernel memory"),
1030 		       sysctl_hw_usermem, 0, NULL, 0,
1031 		       CTL_HW, HW_USERMEM64, CTL_EOL);
1032 }
1033 
1034 #ifdef DEBUG
1035 /*
1036  * Debugging related system variables.
1037  */
1038 struct ctldebug /* debug0, */ /* debug1, */ debug2, debug3, debug4;
1039 struct ctldebug debug5, debug6, debug7, debug8, debug9;
1040 struct ctldebug debug10, debug11, debug12, debug13, debug14;
1041 struct ctldebug debug15, debug16, debug17, debug18, debug19;
1042 static struct ctldebug *debugvars[CTL_DEBUG_MAXID] = {
1043 	&debug0, &debug1, &debug2, &debug3, &debug4,
1044 	&debug5, &debug6, &debug7, &debug8, &debug9,
1045 	&debug10, &debug11, &debug12, &debug13, &debug14,
1046 	&debug15, &debug16, &debug17, &debug18, &debug19,
1047 };
1048 
1049 /*
1050  * this setup routine is a replacement for debug_sysctl()
1051  *
1052  * note that it creates several nodes per defined debug variable
1053  */
1054 SYSCTL_SETUP(sysctl_debug_setup, "sysctl debug subtree setup")
1055 {
1056 	struct ctldebug *cdp;
1057 	char nodename[20];
1058 	int i;
1059 
1060 	/*
1061 	 * two ways here:
1062 	 *
1063 	 * the "old" way (debug.name -> value) which was emulated by
1064 	 * the sysctl(8) binary
1065 	 *
1066 	 * the new way, which the sysctl(8) binary was actually using
1067 
1068 	 node	debug
1069 	 node	debug.0
1070 	 string	debug.0.name
1071 	 int	debug.0.value
1072 	 int	debug.name
1073 
1074 	 */
1075 
1076 	sysctl_createv(clog, 0, NULL, NULL,
1077 		       CTLFLAG_PERMANENT,
1078 		       CTLTYPE_NODE, "debug", NULL,
1079 		       NULL, 0, NULL, 0,
1080 		       CTL_DEBUG, CTL_EOL);
1081 
1082 	for (i = 0; i < CTL_DEBUG_MAXID; i++) {
1083 		cdp = debugvars[i];
1084 		if (cdp->debugname == NULL || cdp->debugvar == NULL)
1085 			continue;
1086 
1087 		snprintf(nodename, sizeof(nodename), "debug%d", i);
1088 		sysctl_createv(clog, 0, NULL, NULL,
1089 			       CTLFLAG_PERMANENT|CTLFLAG_HIDDEN,
1090 			       CTLTYPE_NODE, nodename, NULL,
1091 			       NULL, 0, NULL, 0,
1092 			       CTL_DEBUG, i, CTL_EOL);
1093 		sysctl_createv(clog, 0, NULL, NULL,
1094 			       CTLFLAG_PERMANENT|CTLFLAG_HIDDEN,
1095 			       CTLTYPE_STRING, "name", NULL,
1096 			       /*XXXUNCONST*/
1097 			       NULL, 0, __UNCONST(cdp->debugname), 0,
1098 			       CTL_DEBUG, i, CTL_DEBUG_NAME, CTL_EOL);
1099 		sysctl_createv(clog, 0, NULL, NULL,
1100 			       CTLFLAG_PERMANENT|CTLFLAG_HIDDEN,
1101 			       CTLTYPE_INT, "value", NULL,
1102 			       NULL, 0, cdp->debugvar, 0,
1103 			       CTL_DEBUG, i, CTL_DEBUG_VALUE, CTL_EOL);
1104 		sysctl_createv(clog, 0, NULL, NULL,
1105 			       CTLFLAG_PERMANENT,
1106 			       CTLTYPE_INT, cdp->debugname, NULL,
1107 			       NULL, 0, cdp->debugvar, 0,
1108 			       CTL_DEBUG, CTL_CREATE, CTL_EOL);
1109 	}
1110 }
1111 #endif /* DEBUG */
1112 
1113 /*
1114  * ********************************************************************
1115  * section 2: private node-specific helper routines.
1116  * ********************************************************************
1117  */
1118 
1119 #ifdef DIAGNOSTIC
1120 static int
1121 sysctl_kern_trigger_panic(SYSCTLFN_ARGS)
1122 {
1123 	int newtrig, error;
1124 	struct sysctlnode node;
1125 
1126 	newtrig = 0;
1127 	node = *rnode;
1128 	node.sysctl_data = &newtrig;
1129 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1130 	if (error || newp == NULL)
1131 		return (error);
1132 
1133 	if (newtrig != 0)
1134 		panic("Panic triggered");
1135 
1136 	return (error);
1137 }
1138 #endif
1139 
1140 /*
1141  * sysctl helper routine for kern.maxvnodes.  drain vnodes if
1142  * new value is lower than desiredvnodes and then calls reinit
1143  * routines that needs to adjust to the new value.
1144  */
1145 static int
1146 sysctl_kern_maxvnodes(SYSCTLFN_ARGS)
1147 {
1148 	int error, new_vnodes, old_vnodes;
1149 	struct sysctlnode node;
1150 
1151 	new_vnodes = desiredvnodes;
1152 	node = *rnode;
1153 	node.sysctl_data = &new_vnodes;
1154 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1155 	if (error || newp == NULL)
1156 		return (error);
1157 
1158 	old_vnodes = desiredvnodes;
1159 	desiredvnodes = new_vnodes;
1160 	if (new_vnodes < old_vnodes) {
1161 		error = vfs_drainvnodes(new_vnodes, l);
1162 		if (error) {
1163 			desiredvnodes = old_vnodes;
1164 			return (error);
1165 		}
1166 	}
1167 	vfs_reinit();
1168 	nchreinit();
1169 
1170 	return (0);
1171 }
1172 
1173 /*
1174  * sysctl helper routine for rtc_offset - set time after changes
1175  */
1176 static int
1177 sysctl_kern_rtc_offset(SYSCTLFN_ARGS)
1178 {
1179 	struct timespec ts, delta;
1180 	int error, new_rtc_offset;
1181 	struct sysctlnode node;
1182 
1183 	new_rtc_offset = rtc_offset;
1184 	node = *rnode;
1185 	node.sysctl_data = &new_rtc_offset;
1186 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1187 	if (error || newp == NULL)
1188 		return (error);
1189 
1190 	if (kauth_authorize_system(l->l_cred, KAUTH_SYSTEM_TIME,
1191 	    KAUTH_REQ_SYSTEM_TIME_RTCOFFSET,
1192 	    (void *)(u_long)new_rtc_offset, NULL, NULL))
1193 		return (EPERM);
1194 	if (rtc_offset == new_rtc_offset)
1195 		return (0);
1196 
1197 	/* if we change the offset, adjust the time */
1198 	nanotime(&ts);
1199 	delta.tv_sec = 60 * (new_rtc_offset - rtc_offset);
1200 	delta.tv_nsec = 0;
1201 	timespecadd(&ts, &delta, &ts);
1202 	rtc_offset = new_rtc_offset;
1203 	settime(l->l_proc, &ts);
1204 
1205 	return (0);
1206 }
1207 
1208 /*
1209  * sysctl helper routine for kern.maxproc.  ensures that the new
1210  * values are not too low or too high.
1211  */
1212 static int
1213 sysctl_kern_maxproc(SYSCTLFN_ARGS)
1214 {
1215 	int error, nmaxproc;
1216 	struct sysctlnode node;
1217 
1218 	nmaxproc = maxproc;
1219 	node = *rnode;
1220 	node.sysctl_data = &nmaxproc;
1221 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1222 	if (error || newp == NULL)
1223 		return (error);
1224 
1225 	if (nmaxproc < 0 || nmaxproc >= PID_MAX)
1226 		return (EINVAL);
1227 #ifdef __HAVE_CPU_MAXPROC
1228 	if (nmaxproc > cpu_maxproc())
1229 		return (EINVAL);
1230 #endif
1231 	maxproc = nmaxproc;
1232 
1233 	return (0);
1234 }
1235 
1236 /*
1237  * sysctl helper function for kern.hostid.  the hostid is a long, but
1238  * we export it as an int, so we need to give it a little help.
1239  */
1240 static int
1241 sysctl_kern_hostid(SYSCTLFN_ARGS)
1242 {
1243 	int error, inthostid;
1244 	struct sysctlnode node;
1245 
1246 	inthostid = hostid;  /* XXX assumes sizeof int <= sizeof long */
1247 	node = *rnode;
1248 	node.sysctl_data = &inthostid;
1249 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1250 	if (error || newp == NULL)
1251 		return (error);
1252 
1253 	hostid = (unsigned)inthostid;
1254 
1255 	return (0);
1256 }
1257 
1258 /*
1259  * sysctl helper function for kern.hostname and kern.domainnname.
1260  * resets the relevant recorded length when the underlying name is
1261  * changed.
1262  */
1263 static int
1264 sysctl_setlen(SYSCTLFN_ARGS)
1265 {
1266 	int error;
1267 
1268 	error = sysctl_lookup(SYSCTLFN_CALL(rnode));
1269 	if (error || newp == NULL)
1270 		return (error);
1271 
1272 	switch (rnode->sysctl_num) {
1273 	case KERN_HOSTNAME:
1274 		hostnamelen = strlen((const char*)rnode->sysctl_data);
1275 		break;
1276 	case KERN_DOMAINNAME:
1277 		domainnamelen = strlen((const char*)rnode->sysctl_data);
1278 		break;
1279 	}
1280 
1281 	return (0);
1282 }
1283 
1284 /*
1285  * sysctl helper routine for kern.clockrate.  assembles a struct on
1286  * the fly to be returned to the caller.
1287  */
1288 static int
1289 sysctl_kern_clockrate(SYSCTLFN_ARGS)
1290 {
1291 	struct clockinfo clkinfo;
1292 	struct sysctlnode node;
1293 
1294 	clkinfo.tick = tick;
1295 	clkinfo.tickadj = tickadj;
1296 	clkinfo.hz = hz;
1297 	clkinfo.profhz = profhz;
1298 	clkinfo.stathz = stathz ? stathz : hz;
1299 
1300 	node = *rnode;
1301 	node.sysctl_data = &clkinfo;
1302 	return (sysctl_lookup(SYSCTLFN_CALL(&node)));
1303 }
1304 
1305 
1306 /*
1307  * sysctl helper routine for kern.file pseudo-subtree.
1308  */
1309 static int
1310 sysctl_kern_file(SYSCTLFN_ARGS)
1311 {
1312 	int error;
1313 	size_t buflen;
1314 	struct file *fp;
1315 	char *start, *where;
1316 
1317 	start = where = oldp;
1318 	buflen = *oldlenp;
1319 	if (where == NULL) {
1320 		/*
1321 		 * overestimate by 10 files
1322 		 */
1323 		*oldlenp = sizeof(filehead) + (nfiles + 10) * sizeof(struct file);
1324 		return (0);
1325 	}
1326 
1327 	/*
1328 	 * first dcopyout filehead
1329 	 */
1330 	if (buflen < sizeof(filehead)) {
1331 		*oldlenp = 0;
1332 		return (0);
1333 	}
1334 	error = dcopyout(l, &filehead, where, sizeof(filehead));
1335 	if (error)
1336 		return (error);
1337 	buflen -= sizeof(filehead);
1338 	where += sizeof(filehead);
1339 
1340 	/*
1341 	 * followed by an array of file structures
1342 	 */
1343 	LIST_FOREACH(fp, &filehead, f_list) {
1344 		if (kauth_authorize_generic(l->l_cred,
1345 		    KAUTH_GENERIC_CANSEE, fp->f_cred) != 0)
1346 			continue;
1347 		if (buflen < sizeof(struct file)) {
1348 			*oldlenp = where - start;
1349 			return (ENOMEM);
1350 		}
1351 		error = dcopyout(l, fp, where, sizeof(struct file));
1352 		if (error)
1353 			return (error);
1354 		buflen -= sizeof(struct file);
1355 		where += sizeof(struct file);
1356 	}
1357 	*oldlenp = where - start;
1358 	return (0);
1359 }
1360 
1361 /*
1362  * sysctl helper routine for kern.autonicetime and kern.autoniceval.
1363  * asserts that the assigned value is in the correct range.
1364  */
1365 static int
1366 sysctl_kern_autonice(SYSCTLFN_ARGS)
1367 {
1368 	int error, t = 0;
1369 	struct sysctlnode node;
1370 
1371 	node = *rnode;
1372 	t = *(int*)node.sysctl_data;
1373 	node.sysctl_data = &t;
1374 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1375 	if (error || newp == NULL)
1376 		return (error);
1377 
1378 	switch (node.sysctl_num) {
1379 	case KERN_AUTONICETIME:
1380 		if (t >= 0)
1381 			autonicetime = t;
1382 		break;
1383 	case KERN_AUTONICEVAL:
1384 		if (t < PRIO_MIN)
1385 			t = PRIO_MIN;
1386 		else if (t > PRIO_MAX)
1387 			t = PRIO_MAX;
1388 		autoniceval = t;
1389 		break;
1390 	}
1391 
1392 	return (0);
1393 }
1394 
1395 /*
1396  * sysctl helper routine for kern.msgbufsize and kern.msgbuf.  for the
1397  * former it merely checks the message buffer is set up.  for the latter,
1398  * it also copies out the data if necessary.
1399  */
1400 static int
1401 sysctl_msgbuf(SYSCTLFN_ARGS)
1402 {
1403 	char *where = oldp;
1404 	size_t len, maxlen;
1405 	long beg, end;
1406 	int error;
1407 
1408 	if (!msgbufenabled || msgbufp->msg_magic != MSG_MAGIC) {
1409 		msgbufenabled = 0;
1410 		return (ENXIO);
1411 	}
1412 
1413 	switch (rnode->sysctl_num) {
1414 	case KERN_MSGBUFSIZE: {
1415 		struct sysctlnode node = *rnode;
1416 		int msg_bufs = (int)msgbufp->msg_bufs;
1417 		node.sysctl_data = &msg_bufs;
1418 		return (sysctl_lookup(SYSCTLFN_CALL(&node)));
1419 	}
1420 	case KERN_MSGBUF:
1421 		break;
1422 	default:
1423 		return (EOPNOTSUPP);
1424 	}
1425 
1426 	if (newp != NULL)
1427 		return (EPERM);
1428 
1429         if (oldp == NULL) {
1430 		/* always return full buffer size */
1431 		*oldlenp = msgbufp->msg_bufs;
1432 		return (0);
1433         }
1434 
1435 	error = 0;
1436 	maxlen = MIN(msgbufp->msg_bufs, *oldlenp);
1437 
1438 	/*
1439 	 * First, copy from the write pointer to the end of
1440 	 * message buffer.
1441 	 */
1442 	beg = msgbufp->msg_bufx;
1443 	end = msgbufp->msg_bufs;
1444 	while (maxlen > 0) {
1445 		len = MIN(end - beg, maxlen);
1446 		if (len == 0)
1447 			break;
1448 		error = dcopyout(l, &msgbufp->msg_bufc[beg], where, len);
1449 		if (error)
1450 			break;
1451 		where += len;
1452 		maxlen -= len;
1453 
1454 		/*
1455 		 * ... then, copy from the beginning of message buffer to
1456 		 * the write pointer.
1457 		 */
1458 		beg = 0;
1459 		end = msgbufp->msg_bufx;
1460 	}
1461 
1462 	return (error);
1463 }
1464 
1465 /*
1466  * sysctl helper routine for kern.defcorename.  in the case of a new
1467  * string being assigned, check that it's not a zero-length string.
1468  * (XXX the check in -current doesn't work, but do we really care?)
1469  */
1470 static int
1471 sysctl_kern_defcorename(SYSCTLFN_ARGS)
1472 {
1473 	int error;
1474 	char *newcorename;
1475 	struct sysctlnode node;
1476 
1477 	newcorename = PNBUF_GET();
1478 	node = *rnode;
1479 	node.sysctl_data = &newcorename[0];
1480 	memcpy(node.sysctl_data, rnode->sysctl_data, MAXPATHLEN);
1481 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1482 	if (error || newp == NULL) {
1483 		goto done;
1484 	}
1485 
1486 	/*
1487 	 * when sysctl_lookup() deals with a string, it's guaranteed
1488 	 * to come back nul terminated.  so there.  :)
1489 	 */
1490 	if (strlen(newcorename) == 0) {
1491 		error = EINVAL;
1492 	} else {
1493 		memcpy(rnode->sysctl_data, node.sysctl_data, MAXPATHLEN);
1494 		error = 0;
1495 	}
1496 done:
1497 	PNBUF_PUT(newcorename);
1498 	return error;
1499 }
1500 
1501 /*
1502  * sysctl helper routine for kern.cp_time node.  adds up cpu time
1503  * across all cpus.
1504  */
1505 static int
1506 sysctl_kern_cptime(SYSCTLFN_ARGS)
1507 {
1508 	struct sysctlnode node = *rnode;
1509 
1510 #ifndef MULTIPROCESSOR
1511 
1512 	if (namelen == 1) {
1513 		if (name[0] != 0)
1514 			return (ENOENT);
1515 		/*
1516 		 * you're allowed to ask for the zero'th processor
1517 		 */
1518 		name++;
1519 		namelen--;
1520 	}
1521 	node.sysctl_data = curcpu()->ci_schedstate.spc_cp_time;
1522 	node.sysctl_size = sizeof(curcpu()->ci_schedstate.spc_cp_time);
1523 	return (sysctl_lookup(SYSCTLFN_CALL(&node)));
1524 
1525 #else /* MULTIPROCESSOR */
1526 
1527 	uint64_t *cp_time = NULL;
1528 	int error, n = sysctl_ncpus(), i;
1529 	struct cpu_info *ci;
1530 	CPU_INFO_ITERATOR cii;
1531 
1532 	/*
1533 	 * if you specifically pass a buffer that is the size of the
1534 	 * sum, or if you are probing for the size, you get the "sum"
1535 	 * of cp_time (and the size thereof) across all processors.
1536 	 *
1537 	 * alternately, you can pass an additional mib number and get
1538 	 * cp_time for that particular processor.
1539 	 */
1540 	switch (namelen) {
1541 	case 0:
1542 	    	if (*oldlenp == sizeof(uint64_t) * CPUSTATES || oldp == NULL) {
1543 			node.sysctl_size = sizeof(uint64_t) * CPUSTATES;
1544 			n = -1; /* SUM */
1545 		}
1546 		else {
1547 			node.sysctl_size = n * sizeof(uint64_t) * CPUSTATES;
1548 			n = -2; /* ALL */
1549 		}
1550 		break;
1551 	case 1:
1552 		if (name[0] < 0 || name[0] >= n)
1553 			return (ENOENT); /* ENOSUCHPROCESSOR */
1554 		node.sysctl_size = sizeof(uint64_t) * CPUSTATES;
1555 		n = name[0];
1556 		/*
1557 		 * adjust these so that sysctl_lookup() will be happy
1558 		 */
1559 		name++;
1560 		namelen--;
1561 		break;
1562 	default:
1563 		return (EINVAL);
1564 	}
1565 
1566 	cp_time = malloc(node.sysctl_size, M_TEMP, M_WAITOK|M_CANFAIL);
1567 	if (cp_time == NULL)
1568 		return (ENOMEM);
1569 	node.sysctl_data = cp_time;
1570 	memset(cp_time, 0, node.sysctl_size);
1571 
1572 	for (CPU_INFO_FOREACH(cii, ci)) {
1573 		if (n <= 0)
1574 			for (i = 0; i < CPUSTATES; i++)
1575 				cp_time[i] += ci->ci_schedstate.spc_cp_time[i];
1576 		/*
1577 		 * if a specific processor was requested and we just
1578 		 * did it, we're done here
1579 		 */
1580 		if (n == 0)
1581 			break;
1582 		/*
1583 		 * if doing "all", skip to next cp_time set for next processor
1584 		 */
1585 		if (n == -2)
1586 			cp_time += CPUSTATES;
1587 		/*
1588 		 * if we're doing a specific processor, we're one
1589 		 * processor closer
1590 		 */
1591 		if (n > 0)
1592 			n--;
1593 	}
1594 
1595 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1596 	free(node.sysctl_data, M_TEMP);
1597 	return (error);
1598 
1599 #endif /* MULTIPROCESSOR */
1600 }
1601 
1602 #if defined(SYSVMSG) || defined(SYSVSEM) || defined(SYSVSHM)
1603 /*
1604  * sysctl helper routine for kern.sysvipc_info subtree.
1605  */
1606 
1607 #define	FILL_PERM(src, dst) do { \
1608 	(dst)._key = (src)._key; \
1609 	(dst).uid = (src).uid; \
1610 	(dst).gid = (src).gid; \
1611 	(dst).cuid = (src).cuid; \
1612 	(dst).cgid = (src).cgid; \
1613 	(dst).mode = (src).mode; \
1614 	(dst)._seq = (src)._seq; \
1615 } while (/*CONSTCOND*/ 0);
1616 #define	FILL_MSG(src, dst) do { \
1617 	FILL_PERM((src).msg_perm, (dst).msg_perm); \
1618 	(dst).msg_qnum = (src).msg_qnum; \
1619 	(dst).msg_qbytes = (src).msg_qbytes; \
1620 	(dst)._msg_cbytes = (src)._msg_cbytes; \
1621 	(dst).msg_lspid = (src).msg_lspid; \
1622 	(dst).msg_lrpid = (src).msg_lrpid; \
1623 	(dst).msg_stime = (src).msg_stime; \
1624 	(dst).msg_rtime = (src).msg_rtime; \
1625 	(dst).msg_ctime = (src).msg_ctime; \
1626 } while (/*CONSTCOND*/ 0)
1627 #define	FILL_SEM(src, dst) do { \
1628 	FILL_PERM((src).sem_perm, (dst).sem_perm); \
1629 	(dst).sem_nsems = (src).sem_nsems; \
1630 	(dst).sem_otime = (src).sem_otime; \
1631 	(dst).sem_ctime = (src).sem_ctime; \
1632 } while (/*CONSTCOND*/ 0)
1633 #define	FILL_SHM(src, dst) do { \
1634 	FILL_PERM((src).shm_perm, (dst).shm_perm); \
1635 	(dst).shm_segsz = (src).shm_segsz; \
1636 	(dst).shm_lpid = (src).shm_lpid; \
1637 	(dst).shm_cpid = (src).shm_cpid; \
1638 	(dst).shm_atime = (src).shm_atime; \
1639 	(dst).shm_dtime = (src).shm_dtime; \
1640 	(dst).shm_ctime = (src).shm_ctime; \
1641 	(dst).shm_nattch = (src).shm_nattch; \
1642 } while (/*CONSTCOND*/ 0)
1643 
1644 static int
1645 sysctl_kern_sysvipc(SYSCTLFN_ARGS)
1646 {
1647 	void *where = oldp;
1648 	size_t *sizep = oldlenp;
1649 #ifdef SYSVMSG
1650 	struct msg_sysctl_info *msgsi = NULL;
1651 #endif
1652 #ifdef SYSVSEM
1653 	struct sem_sysctl_info *semsi = NULL;
1654 #endif
1655 #ifdef SYSVSHM
1656 	struct shm_sysctl_info *shmsi = NULL;
1657 #endif
1658 	size_t infosize, dssize, tsize, buflen;
1659 	void *bf = NULL;
1660 	char *start;
1661 	int32_t nds;
1662 	int i, error, ret;
1663 
1664 	if (namelen != 1)
1665 		return (EINVAL);
1666 
1667 	start = where;
1668 	buflen = *sizep;
1669 
1670 	switch (*name) {
1671 	case KERN_SYSVIPC_MSG_INFO:
1672 #ifdef SYSVMSG
1673 		infosize = sizeof(msgsi->msginfo);
1674 		nds = msginfo.msgmni;
1675 		dssize = sizeof(msgsi->msgids[0]);
1676 		break;
1677 #else
1678 		return (EINVAL);
1679 #endif
1680 	case KERN_SYSVIPC_SEM_INFO:
1681 #ifdef SYSVSEM
1682 		infosize = sizeof(semsi->seminfo);
1683 		nds = seminfo.semmni;
1684 		dssize = sizeof(semsi->semids[0]);
1685 		break;
1686 #else
1687 		return (EINVAL);
1688 #endif
1689 	case KERN_SYSVIPC_SHM_INFO:
1690 #ifdef SYSVSHM
1691 		infosize = sizeof(shmsi->shminfo);
1692 		nds = shminfo.shmmni;
1693 		dssize = sizeof(shmsi->shmids[0]);
1694 		break;
1695 #else
1696 		return (EINVAL);
1697 #endif
1698 	default:
1699 		return (EINVAL);
1700 	}
1701 	/*
1702 	 * Round infosize to 64 bit boundary if requesting more than just
1703 	 * the info structure or getting the total data size.
1704 	 */
1705 	if (where == NULL || *sizep > infosize)
1706 		infosize = ((infosize + 7) / 8) * 8;
1707 	tsize = infosize + nds * dssize;
1708 
1709 	/* Return just the total size required. */
1710 	if (where == NULL) {
1711 		*sizep = tsize;
1712 		return (0);
1713 	}
1714 
1715 	/* Not enough room for even the info struct. */
1716 	if (buflen < infosize) {
1717 		*sizep = 0;
1718 		return (ENOMEM);
1719 	}
1720 	bf = malloc(min(tsize, buflen), M_TEMP, M_WAITOK);
1721 	memset(bf, 0, min(tsize, buflen));
1722 
1723 	switch (*name) {
1724 #ifdef SYSVMSG
1725 	case KERN_SYSVIPC_MSG_INFO:
1726 		msgsi = (struct msg_sysctl_info *)bf;
1727 		msgsi->msginfo = msginfo;
1728 		break;
1729 #endif
1730 #ifdef SYSVSEM
1731 	case KERN_SYSVIPC_SEM_INFO:
1732 		semsi = (struct sem_sysctl_info *)bf;
1733 		semsi->seminfo = seminfo;
1734 		break;
1735 #endif
1736 #ifdef SYSVSHM
1737 	case KERN_SYSVIPC_SHM_INFO:
1738 		shmsi = (struct shm_sysctl_info *)bf;
1739 		shmsi->shminfo = shminfo;
1740 		break;
1741 #endif
1742 	}
1743 	buflen -= infosize;
1744 
1745 	ret = 0;
1746 	if (buflen > 0) {
1747 		/* Fill in the IPC data structures.  */
1748 		for (i = 0; i < nds; i++) {
1749 			if (buflen < dssize) {
1750 				ret = ENOMEM;
1751 				break;
1752 			}
1753 			switch (*name) {
1754 #ifdef SYSVMSG
1755 			case KERN_SYSVIPC_MSG_INFO:
1756 				FILL_MSG(msqids[i], msgsi->msgids[i]);
1757 				break;
1758 #endif
1759 #ifdef SYSVSEM
1760 			case KERN_SYSVIPC_SEM_INFO:
1761 				FILL_SEM(sema[i], semsi->semids[i]);
1762 				break;
1763 #endif
1764 #ifdef SYSVSHM
1765 			case KERN_SYSVIPC_SHM_INFO:
1766 				FILL_SHM(shmsegs[i], shmsi->shmids[i]);
1767 				break;
1768 #endif
1769 			}
1770 			buflen -= dssize;
1771 		}
1772 	}
1773 	*sizep -= buflen;
1774 	error = dcopyout(l, bf, start, *sizep);
1775 	/* If dcopyout succeeded, use return code set earlier. */
1776 	if (error == 0)
1777 		error = ret;
1778 	if (bf)
1779 		free(bf, M_TEMP);
1780 	return (error);
1781 }
1782 
1783 #undef FILL_PERM
1784 #undef FILL_MSG
1785 #undef FILL_SEM
1786 #undef FILL_SHM
1787 
1788 #endif /* defined(SYSVMSG) || defined(SYSVSEM) || defined(SYSVSHM) */
1789 
1790 #if NPTY > 0
1791 /*
1792  * sysctl helper routine for kern.maxptys.  ensures that any new value
1793  * is acceptable to the pty subsystem.
1794  */
1795 static int
1796 sysctl_kern_maxptys(SYSCTLFN_ARGS)
1797 {
1798 	int pty_maxptys(int, int);		/* defined in kern/tty_pty.c */
1799 	int error, xmax;
1800 	struct sysctlnode node;
1801 
1802 	/* get current value of maxptys */
1803 	xmax = pty_maxptys(0, 0);
1804 
1805 	node = *rnode;
1806 	node.sysctl_data = &xmax;
1807 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1808 	if (error || newp == NULL)
1809 		return (error);
1810 
1811 	if (xmax != pty_maxptys(xmax, 1))
1812 		return (EINVAL);
1813 
1814 	return (0);
1815 }
1816 #endif /* NPTY > 0 */
1817 
1818 /*
1819  * sysctl helper routine for kern.sbmax.  basically just ensures that
1820  * any new value is not too small.
1821  */
1822 static int
1823 sysctl_kern_sbmax(SYSCTLFN_ARGS)
1824 {
1825 	int error, new_sbmax;
1826 	struct sysctlnode node;
1827 
1828 	new_sbmax = sb_max;
1829 	node = *rnode;
1830 	node.sysctl_data = &new_sbmax;
1831 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1832 	if (error || newp == NULL)
1833 		return (error);
1834 
1835 	error = sb_max_set(new_sbmax);
1836 
1837 	return (error);
1838 }
1839 
1840 /*
1841  * sysctl helper routine for kern.urandom node.  picks a random number
1842  * for you.
1843  */
1844 static int
1845 sysctl_kern_urnd(SYSCTLFN_ARGS)
1846 {
1847 #if NRND > 0
1848 	int v;
1849 
1850 	if (rnd_extract_data(&v, sizeof(v), RND_EXTRACT_ANY) == sizeof(v)) {
1851 		struct sysctlnode node = *rnode;
1852 		node.sysctl_data = &v;
1853 		return (sysctl_lookup(SYSCTLFN_CALL(&node)));
1854 	}
1855 	else
1856 		return (EIO);	/*XXX*/
1857 #else
1858 	return (EOPNOTSUPP);
1859 #endif
1860 }
1861 
1862 /*
1863  * sysctl helper routine for kern.arandom node.  picks a random number
1864  * for you.
1865  */
1866 static int
1867 sysctl_kern_arnd(SYSCTLFN_ARGS)
1868 {
1869 #if NRND > 0
1870 	int error;
1871 	void *v;
1872 	struct sysctlnode node = *rnode;
1873 
1874 	if (*oldlenp == 0)
1875 		return 0;
1876 	if (*oldlenp > 8192)
1877 		return E2BIG;
1878 
1879 	v = malloc(*oldlenp, M_TEMP, M_WAITOK);
1880 
1881 	arc4randbytes(v, *oldlenp);
1882 	node.sysctl_data = v;
1883 	node.sysctl_size = *oldlenp;
1884 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1885 	free(v, M_TEMP);
1886 	return error;
1887 #else
1888 	return (EOPNOTSUPP);
1889 #endif
1890 }
1891 /*
1892  * sysctl helper routine to do kern.lwp.* work.
1893  */
1894 static int
1895 sysctl_kern_lwp(SYSCTLFN_ARGS)
1896 {
1897 	struct kinfo_lwp klwp;
1898 	struct proc *p;
1899 	struct lwp *l2;
1900 	char *where, *dp;
1901 	int pid, elem_size, elem_count;
1902 	int buflen, needed, error;
1903 
1904 	if (namelen == 1 && name[0] == CTL_QUERY)
1905 		return (sysctl_query(SYSCTLFN_CALL(rnode)));
1906 
1907 	dp = where = oldp;
1908 	buflen = where != NULL ? *oldlenp : 0;
1909 	error = needed = 0;
1910 
1911 	if (newp != NULL || namelen != 3)
1912 		return (EINVAL);
1913 	pid = name[0];
1914 	elem_size = name[1];
1915 	elem_count = name[2];
1916 
1917 	p = pfind(pid);
1918 	if (p == NULL)
1919 		return (ESRCH);
1920 	LIST_FOREACH(l2, &p->p_lwps, l_sibling) {
1921 		if (buflen >= elem_size && elem_count > 0) {
1922 			fill_lwp(l2, &klwp);
1923 			/*
1924 			 * Copy out elem_size, but not larger than
1925 			 * the size of a struct kinfo_proc2.
1926 			 */
1927 			error = dcopyout(l, &klwp, dp,
1928 			    min(sizeof(klwp), elem_size));
1929 			if (error)
1930 				goto cleanup;
1931 			dp += elem_size;
1932 			buflen -= elem_size;
1933 			elem_count--;
1934 		}
1935 		needed += elem_size;
1936 	}
1937 
1938 	if (where != NULL) {
1939 		*oldlenp = dp - where;
1940 		if (needed > *oldlenp)
1941 			return (ENOMEM);
1942 	} else {
1943 		needed += KERN_LWPSLOP;
1944 		*oldlenp = needed;
1945 	}
1946 	return (0);
1947  cleanup:
1948 	return (error);
1949 }
1950 
1951 /*
1952  * sysctl helper routine for kern.forkfsleep node.  ensures that the
1953  * given value is not too large or two small, and is at least one
1954  * timer tick if not zero.
1955  */
1956 static int
1957 sysctl_kern_forkfsleep(SYSCTLFN_ARGS)
1958 {
1959 	/* userland sees value in ms, internally is in ticks */
1960 	extern int forkfsleep;		/* defined in kern/kern_fork.c */
1961 	int error, timo, lsleep;
1962 	struct sysctlnode node;
1963 
1964 	lsleep = forkfsleep * 1000 / hz;
1965 	node = *rnode;
1966 	node.sysctl_data = &lsleep;
1967 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1968 	if (error || newp == NULL)
1969 		return (error);
1970 
1971 	/* refuse negative values, and overly 'long time' */
1972 	if (lsleep < 0 || lsleep > MAXSLP * 1000)
1973 		return (EINVAL);
1974 
1975 	timo = mstohz(lsleep);
1976 
1977 	/* if the interval is >0 ms && <1 tick, use 1 tick */
1978 	if (lsleep != 0 && timo == 0)
1979 		forkfsleep = 1;
1980 	else
1981 		forkfsleep = timo;
1982 
1983 	return (0);
1984 }
1985 
1986 /*
1987  * sysctl helper routine for kern.root_partition
1988  */
1989 static int
1990 sysctl_kern_root_partition(SYSCTLFN_ARGS)
1991 {
1992 	int rootpart = DISKPART(rootdev);
1993 	struct sysctlnode node = *rnode;
1994 
1995 	node.sysctl_data = &rootpart;
1996 	return (sysctl_lookup(SYSCTLFN_CALL(&node)));
1997 }
1998 
1999 /*
2000  * sysctl helper function for kern.drivers
2001  */
2002 static int
2003 sysctl_kern_drivers(SYSCTLFN_ARGS)
2004 {
2005 	int error;
2006 	size_t buflen;
2007 	struct kinfo_drivers kd;
2008 	char *start, *where;
2009 	const char *dname;
2010 	int i;
2011 	extern struct devsw_conv *devsw_conv;
2012 	extern int max_devsw_convs;
2013 
2014 	if (newp != NULL || namelen != 0)
2015 		return (EINVAL);
2016 
2017 	start = where = oldp;
2018 	buflen = *oldlenp;
2019 	if (where == NULL) {
2020 		*oldlenp = max_devsw_convs * sizeof kd;
2021 		return 0;
2022 	}
2023 
2024 	/*
2025 	 * An array of kinfo_drivers structures
2026 	 */
2027 	error = 0;
2028 	for (i = 0; i < max_devsw_convs; i++) {
2029 		dname = devsw_conv[i].d_name;
2030 		if (dname == NULL)
2031 			continue;
2032 		if (buflen < sizeof kd) {
2033 			error = ENOMEM;
2034 			break;
2035 		}
2036 		memset(&kd, 0, sizeof(kd));
2037 		kd.d_bmajor = devsw_conv[i].d_bmajor;
2038 		kd.d_cmajor = devsw_conv[i].d_cmajor;
2039 		strlcpy(kd.d_name, dname, sizeof kd.d_name);
2040 		error = dcopyout(l, &kd, where, sizeof kd);
2041 		if (error != 0)
2042 			break;
2043 		buflen -= sizeof kd;
2044 		where += sizeof kd;
2045 	}
2046 	*oldlenp = where - start;
2047 	return error;
2048 }
2049 
2050 /*
2051  * sysctl helper function for kern.file2
2052  */
2053 static int
2054 sysctl_kern_file2(SYSCTLFN_ARGS)
2055 {
2056 	struct proc *p;
2057 	struct file *fp;
2058 	struct filedesc *fd;
2059 	struct kinfo_file kf;
2060 	char *dp;
2061 	u_int i, op;
2062 	size_t len, needed, elem_size, out_size;
2063 	int error, arg, elem_count;
2064 
2065 	if (namelen == 1 && name[0] == CTL_QUERY)
2066 		return (sysctl_query(SYSCTLFN_CALL(rnode)));
2067 
2068 	if (namelen != 4)
2069 		return (EINVAL);
2070 
2071 	error = 0;
2072 	dp = oldp;
2073 	len = (oldp != NULL) ? *oldlenp : 0;
2074 	op = name[0];
2075 	arg = name[1];
2076 	elem_size = name[2];
2077 	elem_count = name[3];
2078 	out_size = MIN(sizeof(kf), elem_size);
2079 	needed = 0;
2080 
2081 	if (elem_size < 1 || elem_count < 0)
2082 		return (EINVAL);
2083 
2084 	switch (op) {
2085 	case KERN_FILE_BYFILE:
2086 		/*
2087 		 * doesn't use arg so it must be zero
2088 		 */
2089 		if (arg != 0)
2090 			return (EINVAL);
2091 		LIST_FOREACH(fp, &filehead, f_list) {
2092 			if (kauth_authorize_generic(l->l_cred,
2093 			    KAUTH_GENERIC_CANSEE, fp->f_cred) != 0)
2094 				continue;
2095 			if (len >= elem_size && elem_count > 0) {
2096 				fill_file(&kf, fp, NULL, 0);
2097 				error = dcopyout(l, &kf, dp, out_size);
2098 				if (error)
2099 					break;
2100 				dp += elem_size;
2101 				len -= elem_size;
2102 			}
2103 			if (elem_count > 0) {
2104 				needed += elem_size;
2105 				if (elem_count != INT_MAX)
2106 					elem_count--;
2107 			}
2108 		}
2109 		break;
2110 	case KERN_FILE_BYPID:
2111 		if (arg < -1)
2112 			/* -1 means all processes */
2113 			return (EINVAL);
2114 		proclist_lock_read();
2115 		PROCLIST_FOREACH(p, &allproc) {
2116 			if (p->p_stat == SIDL)
2117 				/* skip embryonic processes */
2118 				continue;
2119 			if (kauth_authorize_process(l->l_cred,
2120 			    KAUTH_PROCESS_CANSEE, p, NULL, NULL, NULL) != 0)
2121 				continue;
2122 			if (arg > 0 && p->p_pid != arg)
2123 				/* pick only the one we want */
2124 				/* XXX want 0 to mean "kernel files" */
2125 				continue;
2126 			fd = p->p_fd;
2127 			for (i = 0; i < fd->fd_nfiles; i++) {
2128 				fp = fd->fd_ofiles[i];
2129 				if (fp == NULL || !FILE_IS_USABLE(fp))
2130 					continue;
2131 				if (len >= elem_size && elem_count > 0) {
2132 					fill_file(&kf, fd->fd_ofiles[i],
2133 						  p, i);
2134 					error = dcopyout(l, &kf, dp, out_size);
2135 					if (error)
2136 						break;
2137 					dp += elem_size;
2138 					len -= elem_size;
2139 				}
2140 				if (elem_count > 0) {
2141 					needed += elem_size;
2142 					if (elem_count != INT_MAX)
2143 						elem_count--;
2144 				}
2145 			}
2146 		}
2147 		proclist_unlock_read();
2148 		break;
2149 	default:
2150 		return (EINVAL);
2151 	}
2152 
2153 	if (oldp == NULL)
2154 		needed += KERN_FILESLOP * elem_size;
2155 	*oldlenp = needed;
2156 
2157 	return (error);
2158 }
2159 
2160 static void
2161 fill_file(struct kinfo_file *kp, const struct file *fp, struct proc *p, int i)
2162 {
2163 
2164 	memset(kp, 0, sizeof(*kp));
2165 
2166 	kp->ki_fileaddr =	PTRTOUINT64(fp);
2167 	kp->ki_flag =		fp->f_flag;
2168 	kp->ki_iflags =		fp->f_iflags;
2169 	kp->ki_ftype =		fp->f_type;
2170 	kp->ki_count =		fp->f_count;
2171 	kp->ki_msgcount =	fp->f_msgcount;
2172 	kp->ki_usecount =	fp->f_usecount;
2173 	kp->ki_fucred =		PTRTOUINT64(fp->f_cred);
2174 	kp->ki_fuid =		kauth_cred_geteuid(fp->f_cred);
2175 	kp->ki_fgid =		kauth_cred_getegid(fp->f_cred);
2176 	kp->ki_fops =		PTRTOUINT64(fp->f_ops);
2177 	kp->ki_foffset =	fp->f_offset;
2178 	kp->ki_fdata =		PTRTOUINT64(fp->f_data);
2179 
2180 	/* vnode information to glue this file to something */
2181 	if (fp->f_type == DTYPE_VNODE) {
2182 		struct vnode *vp = (struct vnode *)fp->f_data;
2183 
2184 		kp->ki_vun =	PTRTOUINT64(vp->v_un.vu_socket);
2185 		kp->ki_vsize =	vp->v_size;
2186 		kp->ki_vtype =	vp->v_type;
2187 		kp->ki_vtag =	vp->v_tag;
2188 		kp->ki_vdata =	PTRTOUINT64(vp->v_data);
2189 	}
2190 
2191         /* process information when retrieved via KERN_FILE_BYPID */
2192 	if (p) {
2193 		kp->ki_pid =		p->p_pid;
2194 		kp->ki_fd =		i;
2195 		kp->ki_ofileflags =	p->p_fd->fd_ofileflags[i];
2196 	}
2197 }
2198 
2199 static int
2200 sysctl_doeproc(SYSCTLFN_ARGS)
2201 {
2202 	struct eproc *eproc;
2203 	struct kinfo_proc2 *kproc2;
2204 	struct kinfo_proc *dp;
2205 	struct proc *p;
2206 	const struct proclist_desc *pd;
2207 	char *where, *dp2;
2208 	int type, op, arg;
2209 	u_int elem_size, elem_count;
2210 	size_t buflen, needed;
2211 	int error;
2212 
2213 	if (namelen == 1 && name[0] == CTL_QUERY)
2214 		return (sysctl_query(SYSCTLFN_CALL(rnode)));
2215 
2216 	dp = oldp;
2217 	dp2 = where = oldp;
2218 	buflen = where != NULL ? *oldlenp : 0;
2219 	error = 0;
2220 	needed = 0;
2221 	type = rnode->sysctl_num;
2222 
2223 	if (type == KERN_PROC) {
2224 		if (namelen != 2 && !(namelen == 1 && name[0] == KERN_PROC_ALL))
2225 			return (EINVAL);
2226 		op = name[0];
2227 		if (op != KERN_PROC_ALL)
2228 			arg = name[1];
2229 		else
2230 			arg = 0;		/* Quell compiler warning */
2231 		elem_size = elem_count = 0;	/* Ditto */
2232 	} else {
2233 		if (namelen != 4)
2234 			return (EINVAL);
2235 		op = name[0];
2236 		arg = name[1];
2237 		elem_size = name[2];
2238 		elem_count = name[3];
2239 	}
2240 
2241 	if (type == KERN_PROC) {
2242 		eproc = malloc(sizeof(*eproc), M_TEMP, M_WAITOK);
2243 		kproc2 = NULL;
2244 	} else {
2245 		eproc = NULL;
2246 		kproc2 = malloc(sizeof(*kproc2), M_TEMP, M_WAITOK);
2247 	}
2248 	proclist_lock_read();
2249 
2250 	pd = proclists;
2251 again:
2252 	PROCLIST_FOREACH(p, pd->pd_list) {
2253 		/*
2254 		 * Skip embryonic processes.
2255 		 */
2256 		if (p->p_stat == SIDL)
2257 			continue;
2258 
2259 		if (kauth_authorize_process(l->l_cred,
2260 		    KAUTH_PROCESS_CANSEE, p, NULL, NULL, NULL) != 0)
2261 			continue;
2262 
2263 		/*
2264 		 * TODO - make more efficient (see notes below).
2265 		 * do by session.
2266 		 */
2267 		switch (op) {
2268 
2269 		case KERN_PROC_PID:
2270 			/* could do this with just a lookup */
2271 			if (p->p_pid != (pid_t)arg)
2272 				continue;
2273 			break;
2274 
2275 		case KERN_PROC_PGRP:
2276 			/* could do this by traversing pgrp */
2277 			if (p->p_pgrp->pg_id != (pid_t)arg)
2278 				continue;
2279 			break;
2280 
2281 		case KERN_PROC_SESSION:
2282 			if (p->p_session->s_sid != (pid_t)arg)
2283 				continue;
2284 			break;
2285 
2286 		case KERN_PROC_TTY:
2287 			if (arg == (int) KERN_PROC_TTY_REVOKE) {
2288 				if ((p->p_flag & P_CONTROLT) == 0 ||
2289 				    p->p_session->s_ttyp == NULL ||
2290 				    p->p_session->s_ttyvp != NULL)
2291 					continue;
2292 			} else if ((p->p_flag & P_CONTROLT) == 0 ||
2293 			    p->p_session->s_ttyp == NULL) {
2294 				if ((dev_t)arg != KERN_PROC_TTY_NODEV)
2295 					continue;
2296 			} else if (p->p_session->s_ttyp->t_dev != (dev_t)arg)
2297 				continue;
2298 			break;
2299 
2300 		case KERN_PROC_UID:
2301 			if (kauth_cred_geteuid(p->p_cred) != (uid_t)arg)
2302 				continue;
2303 			break;
2304 
2305 		case KERN_PROC_RUID:
2306 			if (kauth_cred_getuid(p->p_cred) != (uid_t)arg)
2307 				continue;
2308 			break;
2309 
2310 		case KERN_PROC_GID:
2311 			if (kauth_cred_getegid(p->p_cred) != (uid_t)arg)
2312 				continue;
2313 			break;
2314 
2315 		case KERN_PROC_RGID:
2316 			if (kauth_cred_getgid(p->p_cred) != (uid_t)arg)
2317 				continue;
2318 			break;
2319 
2320 		case KERN_PROC_ALL:
2321 			/* allow everything */
2322 			break;
2323 
2324 		default:
2325 			error = EINVAL;
2326 			goto cleanup;
2327 		}
2328 		if (type == KERN_PROC) {
2329 			if (buflen >= sizeof(struct kinfo_proc)) {
2330 				fill_eproc(p, eproc);
2331 				error = dcopyout(l, p, &dp->kp_proc,
2332 				    sizeof(struct proc));
2333 				if (error)
2334 					goto cleanup;
2335 				error = dcopyout(l, eproc, &dp->kp_eproc,
2336 				    sizeof(*eproc));
2337 				if (error)
2338 					goto cleanup;
2339 				dp++;
2340 				buflen -= sizeof(struct kinfo_proc);
2341 			}
2342 			needed += sizeof(struct kinfo_proc);
2343 		} else { /* KERN_PROC2 */
2344 			if (buflen >= elem_size && elem_count > 0) {
2345 				fill_kproc2(p, kproc2);
2346 				/*
2347 				 * Copy out elem_size, but not larger than
2348 				 * the size of a struct kinfo_proc2.
2349 				 */
2350 				error = dcopyout(l, kproc2, dp2,
2351 				    min(sizeof(*kproc2), elem_size));
2352 				if (error)
2353 					goto cleanup;
2354 				dp2 += elem_size;
2355 				buflen -= elem_size;
2356 				elem_count--;
2357 			}
2358 			needed += elem_size;
2359 		}
2360 	}
2361 	pd++;
2362 	if (pd->pd_list != NULL)
2363 		goto again;
2364 	proclist_unlock_read();
2365 
2366 	if (where != NULL) {
2367 		if (type == KERN_PROC)
2368 			*oldlenp = (char *)dp - where;
2369 		else
2370 			*oldlenp = dp2 - where;
2371 		if (needed > *oldlenp) {
2372 			error = ENOMEM;
2373 			goto out;
2374 		}
2375 	} else {
2376 		needed += KERN_PROCSLOP;
2377 		*oldlenp = needed;
2378 	}
2379 	if (kproc2)
2380 		free(kproc2, M_TEMP);
2381 	if (eproc)
2382 		free(eproc, M_TEMP);
2383 	return 0;
2384  cleanup:
2385 	proclist_unlock_read();
2386  out:
2387 	if (kproc2)
2388 		free(kproc2, M_TEMP);
2389 	if (eproc)
2390 		free(eproc, M_TEMP);
2391 	return error;
2392 }
2393 
2394 /*
2395  * sysctl helper routine for kern.proc_args pseudo-subtree.
2396  */
2397 static int
2398 sysctl_kern_proc_args(SYSCTLFN_ARGS)
2399 {
2400 	struct ps_strings pss;
2401 	struct proc *p;
2402 	size_t len, i;
2403 	struct uio auio;
2404 	struct iovec aiov;
2405 	pid_t pid;
2406 	int nargv, type, error;
2407 	char *arg;
2408 	char **argv = NULL;
2409 	char *tmp;
2410 	struct vmspace *vmspace;
2411 	vaddr_t psstr_addr;
2412 	vaddr_t offsetn;
2413 	vaddr_t offsetv;
2414 
2415 	if (namelen == 1 && name[0] == CTL_QUERY)
2416 		return (sysctl_query(SYSCTLFN_CALL(rnode)));
2417 
2418 	if (newp != NULL || namelen != 2)
2419 		return (EINVAL);
2420 	pid = name[0];
2421 	type = name[1];
2422 
2423 	switch (type) {
2424 	case KERN_PROC_ARGV:
2425 	case KERN_PROC_NARGV:
2426 	case KERN_PROC_ENV:
2427 	case KERN_PROC_NENV:
2428 		/* ok */
2429 		break;
2430 	default:
2431 		return (EINVAL);
2432 	}
2433 
2434 	proclist_lock_read();
2435 
2436 	/* check pid */
2437 	if ((p = p_find(pid, PFIND_LOCKED)) == NULL) {
2438 		error = EINVAL;
2439 		goto out_locked;
2440 	}
2441 
2442 	error = kauth_authorize_process(l->l_cred,
2443 	    KAUTH_PROCESS_CANSEE, p, NULL, NULL, NULL);
2444 	if (error) {
2445 		goto out_locked;
2446 	}
2447 
2448 	/* only root or same user change look at the environment */
2449 	if (type == KERN_PROC_ENV || type == KERN_PROC_NENV) {
2450 		if (kauth_authorize_process(l->l_cred, KAUTH_PROCESS_CANSEE,
2451 		     p, NULL, NULL, NULL)) {
2452 				error = EPERM;
2453 				goto out_locked;
2454 		}
2455 	}
2456 
2457 	if (oldp == NULL) {
2458 		if (type == KERN_PROC_NARGV || type == KERN_PROC_NENV)
2459 			*oldlenp = sizeof (int);
2460 		else
2461 			*oldlenp = ARG_MAX;	/* XXX XXX XXX */
2462 		error = 0;
2463 		goto out_locked;
2464 	}
2465 
2466 	/*
2467 	 * Zombies don't have a stack, so we can't read their psstrings.
2468 	 * System processes also don't have a user stack.
2469 	 */
2470 	if (P_ZOMBIE(p) || (p->p_flag & P_SYSTEM) != 0) {
2471 		error = EINVAL;
2472 		goto out_locked;
2473 	}
2474 
2475 	/*
2476 	 * Lock the process down in memory.
2477 	 */
2478 	/* XXXCDC: how should locking work here? */
2479 	if ((p->p_flag & P_WEXIT) || (p->p_vmspace->vm_refcnt < 1)) {
2480 		error = EFAULT;
2481 		goto out_locked;
2482 	}
2483 
2484 	psstr_addr = (vaddr_t)p->p_psstr;
2485 	if (type == KERN_PROC_ARGV || type == KERN_PROC_NARGV) {
2486 		offsetn = p->p_psnargv;
2487 		offsetv = p->p_psargv;
2488 	} else {
2489 		offsetn = p->p_psnenv;
2490 		offsetv = p->p_psenv;
2491 	}
2492 	vmspace = p->p_vmspace;
2493 	vmspace->vm_refcnt++;	/* XXX */
2494 
2495 	proclist_unlock_read();
2496 
2497 	/*
2498 	 * Allocate a temporary buffer to hold the arguments.
2499 	 */
2500 	arg = malloc(PAGE_SIZE, M_TEMP, M_WAITOK);
2501 
2502 	/*
2503 	 * Read in the ps_strings structure.
2504 	 */
2505 	aiov.iov_base = &pss;
2506 	aiov.iov_len = sizeof(pss);
2507 	auio.uio_iov = &aiov;
2508 	auio.uio_iovcnt = 1;
2509 	auio.uio_offset = psstr_addr;
2510 	auio.uio_resid = sizeof(pss);
2511 	auio.uio_rw = UIO_READ;
2512 	UIO_SETUP_SYSSPACE(&auio);
2513 	error = uvm_io(&vmspace->vm_map, &auio);
2514 	if (error)
2515 		goto done;
2516 
2517 	memcpy(&nargv, (char *)&pss + offsetn, sizeof(nargv));
2518 	if (type == KERN_PROC_NARGV || type == KERN_PROC_NENV) {
2519 		error = dcopyout(l, &nargv, oldp, sizeof(nargv));
2520 		*oldlenp = sizeof(nargv);
2521 		goto done;
2522 	}
2523 	/*
2524 	 * Now read the address of the argument vector.
2525 	 */
2526 	switch (type) {
2527 	case KERN_PROC_ARGV:
2528 		/* FALLTHROUGH */
2529 	case KERN_PROC_ENV:
2530 		memcpy(&tmp, (char *)&pss + offsetv, sizeof(tmp));
2531 		break;
2532 	default:
2533 		return (EINVAL);
2534 	}
2535 
2536 #ifdef COMPAT_NETBSD32
2537 	if (p->p_flag & P_32)
2538 		len = sizeof(netbsd32_charp) * nargv;
2539 	else
2540 #endif
2541 		len = sizeof(char *) * nargv;
2542 
2543 	argv = malloc(len, M_TEMP, M_WAITOK);
2544 
2545 	aiov.iov_base = argv;
2546 	aiov.iov_len = len;
2547 	auio.uio_iov = &aiov;
2548 	auio.uio_iovcnt = 1;
2549 	auio.uio_offset = (off_t)(unsigned long)tmp;
2550 	auio.uio_resid = len;
2551 	auio.uio_rw = UIO_READ;
2552 	UIO_SETUP_SYSSPACE(&auio);
2553 	error = uvm_io(&vmspace->vm_map, &auio);
2554 	if (error)
2555 		goto done;
2556 
2557 	/*
2558 	 * Now copy each string.
2559 	 */
2560 	len = 0; /* bytes written to user buffer */
2561 	for (i = 0; i < nargv; i++) {
2562 		int finished = 0;
2563 		vaddr_t base;
2564 		size_t xlen;
2565 		int j;
2566 
2567 #ifdef COMPAT_NETBSD32
2568 		if (p->p_flag & P_32) {
2569 			netbsd32_charp *argv32;
2570 
2571 			argv32 = (netbsd32_charp *)argv;
2572 
2573 			base = (vaddr_t)NETBSD32PTR64(argv32[i]);
2574 		} else
2575 #endif
2576 			base = (vaddr_t)argv[i];
2577 
2578 		while (!finished) {
2579 			xlen = PAGE_SIZE - (base & PAGE_MASK);
2580 
2581 			aiov.iov_base = arg;
2582 			aiov.iov_len = PAGE_SIZE;
2583 			auio.uio_iov = &aiov;
2584 			auio.uio_iovcnt = 1;
2585 			auio.uio_offset = base;
2586 			auio.uio_resid = xlen;
2587 			auio.uio_rw = UIO_READ;
2588 			UIO_SETUP_SYSSPACE(&auio);
2589 			error = uvm_io(&vmspace->vm_map, &auio);
2590 			if (error)
2591 				goto done;
2592 
2593 			/* Look for the end of the string */
2594 			for (j = 0; j < xlen; j++) {
2595 				if (arg[j] == '\0') {
2596 					xlen = j + 1;
2597 					finished = 1;
2598 					break;
2599 				}
2600 			}
2601 
2602 			/* Check for user buffer overflow */
2603 			if (len + xlen > *oldlenp) {
2604 				finished = 1;
2605 				if (len > *oldlenp)
2606 					xlen = 0;
2607 				else
2608 					xlen = *oldlenp - len;
2609 			}
2610 
2611 			/* Copyout the page */
2612 			error = dcopyout(l, arg, (char *)oldp + len, xlen);
2613 			if (error)
2614 				goto done;
2615 
2616 			len += xlen;
2617 			base += xlen;
2618 		}
2619 	}
2620 	*oldlenp = len;
2621 
2622 done:
2623 	if (argv != NULL)
2624 		free(argv, M_TEMP);
2625 
2626 	uvmspace_free(vmspace);
2627 
2628 	free(arg, M_TEMP);
2629 	return error;
2630 
2631 out_locked:
2632 	proclist_unlock_read();
2633 	return error;
2634 }
2635 
2636 /*
2637  * Sysctl helper routine for Verified Exec.
2638  */
2639 #if NVERIEXEC > 0
2640 static int
2641 sysctl_kern_veriexec(SYSCTLFN_ARGS)
2642 {
2643 	int newval, error;
2644 	int *var = NULL, raise_only = 0;
2645 	struct sysctlnode node;
2646 
2647 	node = *rnode;
2648 
2649 	switch (rnode->sysctl_num) {
2650 	case VERIEXEC_STRICT:
2651 		raise_only = 1;
2652 		var = &veriexec_strict;
2653 		break;
2654 	case VERIEXEC_ALGORITHMS:
2655 		node.sysctl_data = veriexec_fp_names;
2656 		node.sysctl_size = strlen(veriexec_fp_names) + 1;
2657 		return (sysctl_lookup(SYSCTLFN_CALL(&node)));
2658 	default:
2659 		return (EINVAL);
2660 	}
2661 
2662 	newval = *var;
2663 
2664 	node.sysctl_data = &newval;
2665 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
2666 	if (error || newp == NULL) {
2667 		return (error);
2668 	}
2669 
2670 	if (raise_only && (newval < *var))
2671 		return (EPERM);
2672 
2673 	*var = newval;
2674 
2675 	return (error);
2676 }
2677 #endif /* NVERIEXEC > 0 */
2678 
2679 static int
2680 sysctl_security_setidcore(SYSCTLFN_ARGS)
2681 {
2682 	int newsize, error;
2683 	struct sysctlnode node;
2684 
2685 	node = *rnode;
2686 	node.sysctl_data = &newsize;
2687 	newsize = *(int *)rnode->sysctl_data;
2688 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
2689 	if (error || newp == NULL)
2690 		return error;
2691 
2692 	if (kauth_authorize_system(l->l_cred, KAUTH_SYSTEM_SETIDCORE,
2693 	    0, NULL, NULL, NULL))
2694 		return (EPERM);
2695 
2696 	*(int *)rnode->sysctl_data = newsize;
2697 
2698 	return 0;
2699 }
2700 
2701 static int
2702 sysctl_security_setidcorename(SYSCTLFN_ARGS)
2703 {
2704 	int error;
2705 	char *newsetidcorename;
2706 	struct sysctlnode node;
2707 
2708 	newsetidcorename = PNBUF_GET();
2709 	node = *rnode;
2710 	node.sysctl_data = newsetidcorename;
2711 	memcpy(node.sysctl_data, rnode->sysctl_data, MAXPATHLEN);
2712 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
2713 	if (error || newp == NULL) {
2714 		goto out;
2715 	}
2716 	if (kauth_authorize_system(l->l_cred, KAUTH_SYSTEM_SETIDCORE,
2717 	    0, NULL, NULL, NULL)) {
2718 		error = EPERM;
2719 		goto out;
2720 	}
2721 	if (strlen(newsetidcorename) == 0) {
2722 		error = EINVAL;
2723 		goto out;
2724 	}
2725 	memcpy(rnode->sysctl_data, node.sysctl_data, MAXPATHLEN);
2726 out:
2727 	PNBUF_PUT(newsetidcorename);
2728 	return error;
2729 }
2730 
2731 /*
2732  * sysctl helper routine for kern.cp_id node.  maps cpus to their
2733  * cpuids.
2734  */
2735 static int
2736 sysctl_kern_cpid(SYSCTLFN_ARGS)
2737 {
2738 	struct sysctlnode node = *rnode;
2739 
2740 #ifndef MULTIPROCESSOR
2741 	uint64_t id;
2742 
2743 	if (namelen == 1) {
2744 		if (name[0] != 0)
2745 			return (ENOENT);
2746 		/*
2747 		 * you're allowed to ask for the zero'th processor
2748 		 */
2749 		name++;
2750 		namelen--;
2751 	}
2752 	node.sysctl_data = &id;
2753 	node.sysctl_size = sizeof(id);
2754 	id = cpu_number();
2755 	return (sysctl_lookup(SYSCTLFN_CALL(&node)));
2756 
2757 #else /* MULTIPROCESSOR */
2758 	uint64_t *cp_id = NULL;
2759 	int error, n = sysctl_ncpus();
2760 	struct cpu_info *ci;
2761 	CPU_INFO_ITERATOR cii;
2762 
2763 	/*
2764 	 * here you may either retrieve a single cpu id or the whole
2765 	 * set.  the size you get back when probing depends on what
2766 	 * you ask for.
2767 	 */
2768 	switch (namelen) {
2769 	case 0:
2770 		node.sysctl_size = n * sizeof(uint64_t);
2771 		n = -2; /* ALL */
2772 		break;
2773 	case 1:
2774 		if (name[0] < 0 || name[0] >= n)
2775 			return (ENOENT); /* ENOSUCHPROCESSOR */
2776 		node.sysctl_size = sizeof(uint64_t);
2777 		n = name[0];
2778 		/*
2779 		 * adjust these so that sysctl_lookup() will be happy
2780 		 */
2781 		name++;
2782 		namelen--;
2783 		break;
2784 	default:
2785 		return (EINVAL);
2786 	}
2787 
2788 	cp_id = malloc(node.sysctl_size, M_TEMP, M_WAITOK|M_CANFAIL);
2789 	if (cp_id == NULL)
2790 		return (ENOMEM);
2791 	node.sysctl_data = cp_id;
2792 	memset(cp_id, 0, node.sysctl_size);
2793 
2794 	for (CPU_INFO_FOREACH(cii, ci)) {
2795 		if (n <= 0)
2796 			cp_id[0] = ci->ci_cpuid;
2797 		/*
2798 		 * if a specific processor was requested and we just
2799 		 * did it, we're done here
2800 		 */
2801 		if (n == 0)
2802 			break;
2803 		/*
2804 		 * if doing "all", skip to next cp_id slot for next processor
2805 		 */
2806 		if (n == -2)
2807 			cp_id++;
2808 		/*
2809 		 * if we're doing a specific processor, we're one
2810 		 * processor closer
2811 		 */
2812 		if (n > 0)
2813 			n--;
2814 	}
2815 
2816 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
2817 	free(node.sysctl_data, M_TEMP);
2818 	return (error);
2819 
2820 #endif /* MULTIPROCESSOR */
2821 }
2822 
2823 /*
2824  * sysctl helper routine for hw.usermem and hw.usermem64.  values are
2825  * calculate on the fly taking into account integer overflow and the
2826  * current wired count.
2827  */
2828 static int
2829 sysctl_hw_usermem(SYSCTLFN_ARGS)
2830 {
2831 	u_int ui;
2832 	u_quad_t uq;
2833 	struct sysctlnode node;
2834 
2835 	node = *rnode;
2836 	switch (rnode->sysctl_num) {
2837 	    case HW_USERMEM:
2838 		if ((ui = physmem - uvmexp.wired) > (UINT_MAX / PAGE_SIZE))
2839 			ui = UINT_MAX;
2840 		else
2841 			ui *= PAGE_SIZE;
2842 		node.sysctl_data = &ui;
2843 		break;
2844 	case HW_USERMEM64:
2845 		uq = (u_quad_t)(physmem - uvmexp.wired) * PAGE_SIZE;
2846 		node.sysctl_data = &uq;
2847 		break;
2848 	default:
2849 		return (EINVAL);
2850 	}
2851 
2852 	return (sysctl_lookup(SYSCTLFN_CALL(&node)));
2853 }
2854 
2855 /*
2856  * sysctl helper routine for kern.cnmagic node.  pulls the old value
2857  * out, encoded, and stuffs the new value in for decoding.
2858  */
2859 static int
2860 sysctl_hw_cnmagic(SYSCTLFN_ARGS)
2861 {
2862 	char magic[CNS_LEN];
2863 	int error;
2864 	struct sysctlnode node;
2865 
2866 	if (oldp)
2867 		cn_get_magic(magic, CNS_LEN);
2868 	node = *rnode;
2869 	node.sysctl_data = &magic[0];
2870 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
2871 	if (error || newp == NULL)
2872 		return (error);
2873 
2874 	return (cn_set_magic(magic));
2875 }
2876 
2877 static int
2878 sysctl_hw_ncpu(SYSCTLFN_ARGS)
2879 {
2880 	int ncpu;
2881 	struct sysctlnode node;
2882 
2883 	ncpu = sysctl_ncpus();
2884 	node = *rnode;
2885 	node.sysctl_data = &ncpu;
2886 
2887 	return (sysctl_lookup(SYSCTLFN_CALL(&node)));
2888 }
2889 
2890 
2891 /*
2892  * ********************************************************************
2893  * section 3: public helper routines that are used for more than one
2894  * node
2895  * ********************************************************************
2896  */
2897 
2898 /*
2899  * sysctl helper routine for the kern.root_device node and some ports'
2900  * machdep.root_device nodes.
2901  */
2902 int
2903 sysctl_root_device(SYSCTLFN_ARGS)
2904 {
2905 	struct sysctlnode node;
2906 
2907 	node = *rnode;
2908 	node.sysctl_data = root_device->dv_xname;
2909 	node.sysctl_size = strlen(root_device->dv_xname) + 1;
2910 	return (sysctl_lookup(SYSCTLFN_CALL(&node)));
2911 }
2912 
2913 /*
2914  * sysctl helper routine for kern.consdev, dependent on the current
2915  * state of the console.  also used for machdep.console_device on some
2916  * ports.
2917  */
2918 int
2919 sysctl_consdev(SYSCTLFN_ARGS)
2920 {
2921 	dev_t consdev;
2922 	struct sysctlnode node;
2923 
2924 	if (cn_tab != NULL)
2925 		consdev = cn_tab->cn_dev;
2926 	else
2927 		consdev = NODEV;
2928 	node = *rnode;
2929 	node.sysctl_data = &consdev;
2930 	node.sysctl_size = sizeof(consdev);
2931 	return (sysctl_lookup(SYSCTLFN_CALL(&node)));
2932 }
2933 
2934 /*
2935  * ********************************************************************
2936  * section 4: support for some helpers
2937  * ********************************************************************
2938  */
2939 
2940 /*
2941  * Fill in a kinfo_proc2 structure for the specified process.
2942  */
2943 static void
2944 fill_kproc2(struct proc *p, struct kinfo_proc2 *ki)
2945 {
2946 	struct tty *tp;
2947 	struct lwp *l;
2948 	struct timeval ut, st;
2949 
2950 	memset(ki, 0, sizeof(*ki));
2951 
2952 	ki->p_paddr = PTRTOUINT64(p);
2953 	ki->p_fd = PTRTOUINT64(p->p_fd);
2954 	ki->p_cwdi = PTRTOUINT64(p->p_cwdi);
2955 	ki->p_stats = PTRTOUINT64(p->p_stats);
2956 	ki->p_limit = PTRTOUINT64(p->p_limit);
2957 	ki->p_vmspace = PTRTOUINT64(p->p_vmspace);
2958 	ki->p_sigacts = PTRTOUINT64(p->p_sigacts);
2959 	ki->p_sess = PTRTOUINT64(p->p_session);
2960 	ki->p_tsess = 0;	/* may be changed if controlling tty below */
2961 	ki->p_ru = PTRTOUINT64(p->p_ru);
2962 
2963 	ki->p_eflag = 0;
2964 	ki->p_exitsig = p->p_exitsig;
2965 	ki->p_flag = p->p_flag;
2966 
2967 	ki->p_pid = p->p_pid;
2968 	if (p->p_pptr)
2969 		ki->p_ppid = p->p_pptr->p_pid;
2970 	else
2971 		ki->p_ppid = 0;
2972 	ki->p_sid = p->p_session->s_sid;
2973 	ki->p__pgid = p->p_pgrp->pg_id;
2974 
2975 	ki->p_tpgid = NO_PGID;	/* may be changed if controlling tty below */
2976 
2977 	ki->p_uid = kauth_cred_geteuid(p->p_cred);
2978 	ki->p_ruid = kauth_cred_getuid(p->p_cred);
2979 	ki->p_gid = kauth_cred_getegid(p->p_cred);
2980 	ki->p_rgid = kauth_cred_getgid(p->p_cred);
2981 	ki->p_svuid = kauth_cred_getsvuid(p->p_cred);
2982 	ki->p_svgid = kauth_cred_getsvgid(p->p_cred);
2983 
2984 	ki->p_ngroups = kauth_cred_ngroups(p->p_cred);
2985 	kauth_cred_getgroups(p->p_cred, ki->p_groups,
2986 	    min(ki->p_ngroups, sizeof(ki->p_groups) / sizeof(ki->p_groups[0])));
2987 
2988 	ki->p_jobc = p->p_pgrp->pg_jobc;
2989 	if ((p->p_flag & P_CONTROLT) && (tp = p->p_session->s_ttyp)) {
2990 		ki->p_tdev = tp->t_dev;
2991 		ki->p_tpgid = tp->t_pgrp ? tp->t_pgrp->pg_id : NO_PGID;
2992 		ki->p_tsess = PTRTOUINT64(tp->t_session);
2993 	} else {
2994 		ki->p_tdev = NODEV;
2995 	}
2996 
2997 	ki->p_estcpu = p->p_estcpu;
2998 	ki->p_rtime_sec = p->p_rtime.tv_sec;
2999 	ki->p_rtime_usec = p->p_rtime.tv_usec;
3000 	ki->p_cpticks = p->p_cpticks;
3001 	ki->p_pctcpu = p->p_pctcpu;
3002 
3003 	ki->p_uticks = p->p_uticks;
3004 	ki->p_sticks = p->p_sticks;
3005 	ki->p_iticks = p->p_iticks;
3006 
3007 	ki->p_tracep = PTRTOUINT64(p->p_tracep);
3008 	ki->p_traceflag = p->p_traceflag;
3009 
3010 
3011 	memcpy(&ki->p_siglist, &p->p_sigctx.ps_siglist, sizeof(ki_sigset_t));
3012 	memcpy(&ki->p_sigmask, &p->p_sigctx.ps_sigmask, sizeof(ki_sigset_t));
3013 	memcpy(&ki->p_sigignore, &p->p_sigctx.ps_sigignore,sizeof(ki_sigset_t));
3014 	memcpy(&ki->p_sigcatch, &p->p_sigctx.ps_sigcatch, sizeof(ki_sigset_t));
3015 
3016 	ki->p_stat = p->p_stat; /* Will likely be overridden by LWP status */
3017 	ki->p_realstat = p->p_stat;
3018 	ki->p_nice = p->p_nice;
3019 
3020 	ki->p_xstat = p->p_xstat;
3021 	ki->p_acflag = p->p_acflag;
3022 
3023 	strncpy(ki->p_comm, p->p_comm,
3024 	    min(sizeof(ki->p_comm), sizeof(p->p_comm)));
3025 
3026 	strncpy(ki->p_login, p->p_session->s_login,
3027 	    min(sizeof ki->p_login - 1, sizeof p->p_session->s_login));
3028 
3029 	ki->p_nlwps = p->p_nlwps;
3030 	ki->p_nrlwps = p->p_nrlwps;
3031 	ki->p_realflag = p->p_flag;
3032 
3033 	if (p->p_stat == SIDL || P_ZOMBIE(p)) {
3034 		ki->p_vm_rssize = 0;
3035 		ki->p_vm_tsize = 0;
3036 		ki->p_vm_dsize = 0;
3037 		ki->p_vm_ssize = 0;
3038 		l = NULL;
3039 	} else {
3040 		struct vmspace *vm = p->p_vmspace;
3041 
3042 		ki->p_vm_rssize = vm_resident_count(vm);
3043 		ki->p_vm_tsize = vm->vm_tsize;
3044 		ki->p_vm_dsize = vm->vm_dsize;
3045 		ki->p_vm_ssize = vm->vm_ssize;
3046 
3047 		/* Pick a "representative" LWP */
3048 		l = proc_representative_lwp(p);
3049 		ki->p_forw = PTRTOUINT64(l->l_forw);
3050 		ki->p_back = PTRTOUINT64(l->l_back);
3051 		ki->p_addr = PTRTOUINT64(l->l_addr);
3052 		ki->p_stat = l->l_stat;
3053 		ki->p_flag |= l->l_flag & P_SHARED;
3054 		ki->p_swtime = l->l_swtime;
3055 		ki->p_slptime = l->l_slptime;
3056 		if (l->l_stat == LSONPROC) {
3057 			KDASSERT(l->l_cpu != NULL);
3058 			ki->p_schedflags = l->l_cpu->ci_schedstate.spc_flags;
3059 		} else
3060 			ki->p_schedflags = 0;
3061 		ki->p_holdcnt = l->l_holdcnt;
3062 		ki->p_priority = l->l_priority;
3063 		ki->p_usrpri = l->l_usrpri;
3064 		if (l->l_wmesg)
3065 			strncpy(ki->p_wmesg, l->l_wmesg, sizeof(ki->p_wmesg));
3066 		ki->p_wchan = PTRTOUINT64(l->l_wchan);
3067 
3068 	}
3069 
3070 	if (p->p_session->s_ttyvp)
3071 		ki->p_eflag |= EPROC_CTTY;
3072 	if (SESS_LEADER(p))
3073 		ki->p_eflag |= EPROC_SLEADER;
3074 
3075 	/* XXX Is this double check necessary? */
3076 	if (P_ZOMBIE(p)) {
3077 		ki->p_uvalid = 0;
3078 	} else {
3079 		ki->p_uvalid = 1;
3080 
3081 		ki->p_ustart_sec = p->p_stats->p_start.tv_sec;
3082 		ki->p_ustart_usec = p->p_stats->p_start.tv_usec;
3083 
3084 		calcru(p, &ut, &st, 0);
3085 		ki->p_uutime_sec = ut.tv_sec;
3086 		ki->p_uutime_usec = ut.tv_usec;
3087 		ki->p_ustime_sec = st.tv_sec;
3088 		ki->p_ustime_usec = st.tv_usec;
3089 
3090 		ki->p_uru_maxrss = p->p_stats->p_ru.ru_maxrss;
3091 		ki->p_uru_ixrss = p->p_stats->p_ru.ru_ixrss;
3092 		ki->p_uru_idrss = p->p_stats->p_ru.ru_idrss;
3093 		ki->p_uru_isrss = p->p_stats->p_ru.ru_isrss;
3094 		ki->p_uru_minflt = p->p_stats->p_ru.ru_minflt;
3095 		ki->p_uru_majflt = p->p_stats->p_ru.ru_majflt;
3096 		ki->p_uru_nswap = p->p_stats->p_ru.ru_nswap;
3097 		ki->p_uru_inblock = p->p_stats->p_ru.ru_inblock;
3098 		ki->p_uru_oublock = p->p_stats->p_ru.ru_oublock;
3099 		ki->p_uru_msgsnd = p->p_stats->p_ru.ru_msgsnd;
3100 		ki->p_uru_msgrcv = p->p_stats->p_ru.ru_msgrcv;
3101 		ki->p_uru_nsignals = p->p_stats->p_ru.ru_nsignals;
3102 		ki->p_uru_nvcsw = p->p_stats->p_ru.ru_nvcsw;
3103 		ki->p_uru_nivcsw = p->p_stats->p_ru.ru_nivcsw;
3104 
3105 		timeradd(&p->p_stats->p_cru.ru_utime,
3106 			 &p->p_stats->p_cru.ru_stime, &ut);
3107 		ki->p_uctime_sec = ut.tv_sec;
3108 		ki->p_uctime_usec = ut.tv_usec;
3109 	}
3110 #ifdef MULTIPROCESSOR
3111 	if (l && l->l_cpu != NULL)
3112 		ki->p_cpuid = l->l_cpu->ci_cpuid;
3113 	else
3114 #endif
3115 		ki->p_cpuid = KI_NOCPU;
3116 }
3117 
3118 /*
3119  * Fill in a kinfo_lwp structure for the specified lwp.
3120  */
3121 static void
3122 fill_lwp(struct lwp *l, struct kinfo_lwp *kl)
3123 {
3124 
3125 	kl->l_forw = PTRTOUINT64(l->l_forw);
3126 	kl->l_back = PTRTOUINT64(l->l_back);
3127 	kl->l_laddr = PTRTOUINT64(l);
3128 	kl->l_addr = PTRTOUINT64(l->l_addr);
3129 	kl->l_stat = l->l_stat;
3130 	kl->l_lid = l->l_lid;
3131 	kl->l_flag = l->l_flag;
3132 
3133 	kl->l_swtime = l->l_swtime;
3134 	kl->l_slptime = l->l_slptime;
3135 	if (l->l_stat == LSONPROC) {
3136 		KDASSERT(l->l_cpu != NULL);
3137 		kl->l_schedflags = l->l_cpu->ci_schedstate.spc_flags;
3138 	} else
3139 		kl->l_schedflags = 0;
3140 	kl->l_holdcnt = l->l_holdcnt;
3141 	kl->l_priority = l->l_priority;
3142 	kl->l_usrpri = l->l_usrpri;
3143 	if (l->l_wmesg)
3144 		strncpy(kl->l_wmesg, l->l_wmesg, sizeof(kl->l_wmesg));
3145 	kl->l_wchan = PTRTOUINT64(l->l_wchan);
3146 #ifdef MULTIPROCESSOR
3147 	if (l->l_cpu != NULL)
3148 		kl->l_cpuid = l->l_cpu->ci_cpuid;
3149 	else
3150 #endif
3151 		kl->l_cpuid = KI_NOCPU;
3152 }
3153 
3154 /*
3155  * Fill in an eproc structure for the specified process.
3156  */
3157 void
3158 fill_eproc(struct proc *p, struct eproc *ep)
3159 {
3160 	struct tty *tp;
3161 	struct lwp *l;
3162 
3163 	ep->e_paddr = p;
3164 	ep->e_sess = p->p_session;
3165 	kauth_cred_topcred(p->p_cred, &ep->e_pcred);
3166 	kauth_cred_toucred(p->p_cred, &ep->e_ucred);
3167 	if (p->p_stat == SIDL || P_ZOMBIE(p)) {
3168 		ep->e_vm.vm_rssize = 0;
3169 		ep->e_vm.vm_tsize = 0;
3170 		ep->e_vm.vm_dsize = 0;
3171 		ep->e_vm.vm_ssize = 0;
3172 		/* ep->e_vm.vm_pmap = XXX; */
3173 	} else {
3174 		struct vmspace *vm = p->p_vmspace;
3175 
3176 		ep->e_vm.vm_rssize = vm_resident_count(vm);
3177 		ep->e_vm.vm_tsize = vm->vm_tsize;
3178 		ep->e_vm.vm_dsize = vm->vm_dsize;
3179 		ep->e_vm.vm_ssize = vm->vm_ssize;
3180 
3181 		/* Pick a "representative" LWP */
3182 		l = proc_representative_lwp(p);
3183 
3184 		if (l->l_wmesg)
3185 			strncpy(ep->e_wmesg, l->l_wmesg, WMESGLEN);
3186 	}
3187 	if (p->p_pptr)
3188 		ep->e_ppid = p->p_pptr->p_pid;
3189 	else
3190 		ep->e_ppid = 0;
3191 	ep->e_pgid = p->p_pgrp->pg_id;
3192 	ep->e_sid = ep->e_sess->s_sid;
3193 	ep->e_jobc = p->p_pgrp->pg_jobc;
3194 	if ((p->p_flag & P_CONTROLT) &&
3195 	    (tp = ep->e_sess->s_ttyp)) {
3196 		ep->e_tdev = tp->t_dev;
3197 		ep->e_tpgid = tp->t_pgrp ? tp->t_pgrp->pg_id : NO_PGID;
3198 		ep->e_tsess = tp->t_session;
3199 	} else
3200 		ep->e_tdev = NODEV;
3201 
3202 	ep->e_xsize = ep->e_xrssize = 0;
3203 	ep->e_xccount = ep->e_xswrss = 0;
3204 	ep->e_flag = ep->e_sess->s_ttyvp ? EPROC_CTTY : 0;
3205 	if (SESS_LEADER(p))
3206 		ep->e_flag |= EPROC_SLEADER;
3207 	strncpy(ep->e_login, ep->e_sess->s_login, MAXLOGNAME);
3208 }
3209