xref: /openbsd-src/sys/kern/kern_sysctl.c (revision 94358d69ee05fa503294e6438e1b1bbf60aa9d02)
1 /*	$OpenBSD: kern_sysctl.c,v 1.419 2023/09/16 09:33:27 mpi Exp $	*/
2 /*	$NetBSD: kern_sysctl.c,v 1.17 1996/05/20 17:49:05 mrg Exp $	*/
3 
4 /*-
5  * Copyright (c) 1982, 1986, 1989, 1993
6  *	The Regents of the University of California.  All rights reserved.
7  *
8  * This code is derived from software contributed to Berkeley by
9  * Mike Karels at Berkeley Software Design, Inc.
10  *
11  * Redistribution and use in source and binary forms, with or without
12  * modification, are permitted provided that the following conditions
13  * are met:
14  * 1. Redistributions of source code must retain the above copyright
15  *    notice, this list of conditions and the following disclaimer.
16  * 2. Redistributions in binary form must reproduce the above copyright
17  *    notice, this list of conditions and the following disclaimer in the
18  *    documentation and/or other materials provided with the distribution.
19  * 3. Neither the name of the University nor the names of its contributors
20  *    may be used to endorse or promote products derived from this software
21  *    without specific prior written permission.
22  *
23  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
24  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
25  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
26  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
27  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
28  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
29  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
30  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
31  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
32  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
33  * SUCH DAMAGE.
34  *
35  *	@(#)kern_sysctl.c	8.4 (Berkeley) 4/14/94
36  */
37 
38 /*
39  * sysctl system call.
40  */
41 
42 #include <sys/param.h>
43 #include <sys/systm.h>
44 #include <sys/kernel.h>
45 #include <sys/malloc.h>
46 #include <sys/pool.h>
47 #include <sys/proc.h>
48 #include <sys/resourcevar.h>
49 #include <sys/signalvar.h>
50 #include <sys/fcntl.h>
51 #include <sys/file.h>
52 #include <sys/filedesc.h>
53 #include <sys/vnode.h>
54 #include <sys/unistd.h>
55 #include <sys/buf.h>
56 #include <sys/clockintr.h>
57 #include <sys/tty.h>
58 #include <sys/disklabel.h>
59 #include <sys/disk.h>
60 #include <sys/sysctl.h>
61 #include <sys/msgbuf.h>
62 #include <sys/vmmeter.h>
63 #include <sys/namei.h>
64 #include <sys/exec.h>
65 #include <sys/mbuf.h>
66 #include <sys/percpu.h>
67 #include <sys/sensors.h>
68 #include <sys/pipe.h>
69 #include <sys/eventvar.h>
70 #include <sys/socketvar.h>
71 #include <sys/socket.h>
72 #include <sys/domain.h>
73 #include <sys/protosw.h>
74 #include <sys/pledge.h>
75 #include <sys/timetc.h>
76 #include <sys/evcount.h>
77 #include <sys/un.h>
78 #include <sys/unpcb.h>
79 #include <sys/sched.h>
80 #include <sys/mount.h>
81 #include <sys/syscallargs.h>
82 #include <sys/wait.h>
83 #include <sys/witness.h>
84 
85 #include <uvm/uvm_extern.h>
86 
87 #include <dev/cons.h>
88 
89 #include <net/route.h>
90 #include <netinet/in.h>
91 #include <netinet/ip.h>
92 #include <netinet/ip_var.h>
93 #include <netinet/in_pcb.h>
94 #include <netinet/ip6.h>
95 #include <netinet/tcp.h>
96 #include <netinet/tcp_timer.h>
97 #include <netinet/tcp_var.h>
98 #include <netinet/udp.h>
99 #include <netinet/udp_var.h>
100 #include <netinet6/ip6_var.h>
101 
102 #ifdef DDB
103 #include <ddb/db_var.h>
104 #endif
105 
106 #ifdef SYSVMSG
107 #include <sys/msg.h>
108 #endif
109 #ifdef SYSVSEM
110 #include <sys/sem.h>
111 #endif
112 #ifdef SYSVSHM
113 #include <sys/shm.h>
114 #endif
115 
116 #include "audio.h"
117 #include "dt.h"
118 #include "pf.h"
119 #include "video.h"
120 
121 extern struct forkstat forkstat;
122 extern struct nchstats nchstats;
123 extern int fscale;
124 extern fixpt_t ccpu;
125 extern long numvnodes;
126 extern int allowdt;
127 extern int audio_record_enable;
128 extern int video_record_enable;
129 extern int autoconf_serial;
130 
131 int allowkmem;
132 
133 int sysctl_diskinit(int, struct proc *);
134 int sysctl_proc_args(int *, u_int, void *, size_t *, struct proc *);
135 int sysctl_proc_cwd(int *, u_int, void *, size_t *, struct proc *);
136 int sysctl_proc_nobroadcastkill(int *, u_int, void *, size_t, void *, size_t *,
137 	struct proc *);
138 int sysctl_proc_vmmap(int *, u_int, void *, size_t *, struct proc *);
139 int sysctl_intrcnt(int *, u_int, void *, size_t *);
140 int sysctl_sensors(int *, u_int, void *, size_t *, void *, size_t);
141 int sysctl_cptime2(int *, u_int, void *, size_t *, void *, size_t);
142 int sysctl_audio(int *, u_int, void *, size_t *, void *, size_t);
143 int sysctl_video(int *, u_int, void *, size_t *, void *, size_t);
144 int sysctl_cpustats(int *, u_int, void *, size_t *, void *, size_t);
145 int sysctl_utc_offset(void *, size_t *, void *, size_t);
146 int sysctl_hwbattery(int *, u_int, void *, size_t *, void *, size_t);
147 
148 void fill_file(struct kinfo_file *, struct file *, struct filedesc *, int,
149     struct vnode *, struct process *, struct proc *, struct socket *, int);
150 void fill_kproc(struct process *, struct kinfo_proc *, struct proc *, int);
151 
152 int (*cpu_cpuspeed)(int *);
153 
154 /*
155  * Lock to avoid too many processes vslocking a large amount of memory
156  * at the same time.
157  */
158 struct rwlock sysctl_lock = RWLOCK_INITIALIZER("sysctllk");
159 struct rwlock sysctl_disklock = RWLOCK_INITIALIZER("sysctldlk");
160 
161 int
162 sys_sysctl(struct proc *p, void *v, register_t *retval)
163 {
164 	struct sys_sysctl_args /* {
165 		syscallarg(const int *) name;
166 		syscallarg(u_int) namelen;
167 		syscallarg(void *) old;
168 		syscallarg(size_t *) oldlenp;
169 		syscallarg(void *) new;
170 		syscallarg(size_t) newlen;
171 	} */ *uap = v;
172 	int error, dolock = 1;
173 	size_t savelen = 0, oldlen = 0;
174 	sysctlfn *fn;
175 	int name[CTL_MAXNAME];
176 
177 	if (SCARG(uap, new) != NULL &&
178 	    (error = suser(p)))
179 		return (error);
180 	/*
181 	 * all top-level sysctl names are non-terminal
182 	 */
183 	if (SCARG(uap, namelen) > CTL_MAXNAME || SCARG(uap, namelen) < 2)
184 		return (EINVAL);
185 	error = copyin(SCARG(uap, name), name,
186 		       SCARG(uap, namelen) * sizeof(int));
187 	if (error)
188 		return (error);
189 
190 	error = pledge_sysctl(p, SCARG(uap, namelen),
191 	    name, SCARG(uap, new));
192 	if (error)
193 		return (error);
194 
195 	switch (name[0]) {
196 	case CTL_KERN:
197 		fn = kern_sysctl;
198 		break;
199 	case CTL_HW:
200 		fn = hw_sysctl;
201 		break;
202 	case CTL_VM:
203 		fn = uvm_sysctl;
204 		break;
205 	case CTL_NET:
206 		fn = net_sysctl;
207 		break;
208 	case CTL_FS:
209 		fn = fs_sysctl;
210 		break;
211 	case CTL_VFS:
212 		fn = vfs_sysctl;
213 		break;
214 	case CTL_MACHDEP:
215 		fn = cpu_sysctl;
216 		break;
217 #ifdef DEBUG_SYSCTL
218 	case CTL_DEBUG:
219 		fn = debug_sysctl;
220 		break;
221 #endif
222 #ifdef DDB
223 	case CTL_DDB:
224 		fn = ddb_sysctl;
225 		break;
226 #endif
227 	default:
228 		return (EOPNOTSUPP);
229 	}
230 
231 	if (SCARG(uap, oldlenp) &&
232 	    (error = copyin(SCARG(uap, oldlenp), &oldlen, sizeof(oldlen))))
233 		return (error);
234 	if (SCARG(uap, old) != NULL) {
235 		if ((error = rw_enter(&sysctl_lock, RW_WRITE|RW_INTR)) != 0)
236 			return (error);
237 		if (dolock) {
238 			if (atop(oldlen) > uvmexp.wiredmax - uvmexp.wired) {
239 				rw_exit_write(&sysctl_lock);
240 				return (ENOMEM);
241 			}
242 			error = uvm_vslock(p, SCARG(uap, old), oldlen,
243 			    PROT_READ | PROT_WRITE);
244 			if (error) {
245 				rw_exit_write(&sysctl_lock);
246 				return (error);
247 			}
248 		}
249 		savelen = oldlen;
250 	}
251 	error = (*fn)(&name[1], SCARG(uap, namelen) - 1, SCARG(uap, old),
252 	    &oldlen, SCARG(uap, new), SCARG(uap, newlen), p);
253 	if (SCARG(uap, old) != NULL) {
254 		if (dolock)
255 			uvm_vsunlock(p, SCARG(uap, old), savelen);
256 		rw_exit_write(&sysctl_lock);
257 	}
258 	if (error)
259 		return (error);
260 	if (SCARG(uap, oldlenp))
261 		error = copyout(&oldlen, SCARG(uap, oldlenp), sizeof(oldlen));
262 	return (error);
263 }
264 
265 /*
266  * Attributes stored in the kernel.
267  */
268 char hostname[MAXHOSTNAMELEN];
269 int hostnamelen;
270 char domainname[MAXHOSTNAMELEN];
271 int domainnamelen;
272 long hostid;
273 char *disknames = NULL;
274 size_t disknameslen;
275 struct diskstats *diskstats = NULL;
276 size_t diskstatslen;
277 int securelevel;
278 
279 /* morally const values reported by sysctl_bounded_arr */
280 static int arg_max = ARG_MAX;
281 static int openbsd = OpenBSD;
282 static int posix_version = _POSIX_VERSION;
283 static int ngroups_max = NGROUPS_MAX;
284 static int int_zero = 0;
285 static int int_one = 1;
286 static int maxpartitions = MAXPARTITIONS;
287 static int raw_part = RAW_PART;
288 
289 extern int somaxconn, sominconn;
290 extern int nosuidcoredump;
291 extern int maxlocksperuid;
292 extern int uvm_wxabort;
293 extern int global_ptrace;
294 
295 const struct sysctl_bounded_args kern_vars[] = {
296 	{KERN_OSREV, &openbsd, SYSCTL_INT_READONLY},
297 	{KERN_MAXVNODES, &maxvnodes, 0, INT_MAX},
298 	{KERN_MAXPROC, &maxprocess, 0, INT_MAX},
299 	{KERN_MAXFILES, &maxfiles, 0, INT_MAX},
300 	{KERN_NFILES, &numfiles, SYSCTL_INT_READONLY},
301 	{KERN_TTYCOUNT, &tty_count, SYSCTL_INT_READONLY},
302 	{KERN_ARGMAX, &arg_max, SYSCTL_INT_READONLY},
303 	{KERN_POSIX1, &posix_version, SYSCTL_INT_READONLY},
304 	{KERN_NGROUPS, &ngroups_max, SYSCTL_INT_READONLY},
305 	{KERN_JOB_CONTROL, &int_one, SYSCTL_INT_READONLY},
306 	{KERN_SAVED_IDS, &int_one, SYSCTL_INT_READONLY},
307 	{KERN_MAXPARTITIONS, &maxpartitions, SYSCTL_INT_READONLY},
308 	{KERN_RAWPARTITION, &raw_part, SYSCTL_INT_READONLY},
309 	{KERN_MAXTHREAD, &maxthread, 0, INT_MAX},
310 	{KERN_NTHREADS, &nthreads, SYSCTL_INT_READONLY},
311 	{KERN_SOMAXCONN, &somaxconn, 0, SHRT_MAX},
312 	{KERN_SOMINCONN, &sominconn, 0, SHRT_MAX},
313 	{KERN_NOSUIDCOREDUMP, &nosuidcoredump, 0, 3},
314 	{KERN_FSYNC, &int_one, SYSCTL_INT_READONLY},
315 	{KERN_SYSVMSG,
316 #ifdef SYSVMSG
317 	 &int_one,
318 #else
319 	 &int_zero,
320 #endif
321 	 SYSCTL_INT_READONLY},
322 	{KERN_SYSVSEM,
323 #ifdef SYSVSEM
324 	 &int_one,
325 #else
326 	 &int_zero,
327 #endif
328 	 SYSCTL_INT_READONLY},
329 	{KERN_SYSVSHM,
330 #ifdef SYSVSHM
331 	 &int_one,
332 #else
333 	 &int_zero,
334 #endif
335 	 SYSCTL_INT_READONLY},
336 	{KERN_FSCALE, &fscale, SYSCTL_INT_READONLY},
337 	{KERN_CCPU, &ccpu, SYSCTL_INT_READONLY},
338 	{KERN_NPROCS, &nprocesses, SYSCTL_INT_READONLY},
339 	{KERN_SPLASSERT, &splassert_ctl, 0, 3},
340 	{KERN_MAXLOCKSPERUID, &maxlocksperuid, 0, INT_MAX},
341 	{KERN_WXABORT, &uvm_wxabort, 0, 1},
342 	{KERN_NETLIVELOCKS, &int_zero, SYSCTL_INT_READONLY},
343 #ifdef PTRACE
344 	{KERN_GLOBAL_PTRACE, &global_ptrace, 0, 1},
345 #endif
346 	{KERN_AUTOCONF_SERIAL, &autoconf_serial, SYSCTL_INT_READONLY},
347 };
348 
349 int
350 kern_sysctl_dirs(int top_name, int *name, u_int namelen,
351     void *oldp, size_t *oldlenp, void *newp, size_t newlen, struct proc *p)
352 {
353 	switch (top_name) {
354 #ifndef SMALL_KERNEL
355 	case KERN_PROC:
356 		return (sysctl_doproc(name, namelen, oldp, oldlenp));
357 	case KERN_PROC_ARGS:
358 		return (sysctl_proc_args(name, namelen, oldp, oldlenp, p));
359 	case KERN_PROC_CWD:
360 		return (sysctl_proc_cwd(name, namelen, oldp, oldlenp, p));
361 	case KERN_PROC_NOBROADCASTKILL:
362 		return (sysctl_proc_nobroadcastkill(name, namelen,
363 		     newp, newlen, oldp, oldlenp, p));
364 	case KERN_PROC_VMMAP:
365 		return (sysctl_proc_vmmap(name, namelen, oldp, oldlenp, p));
366 	case KERN_FILE:
367 		return (sysctl_file(name, namelen, oldp, oldlenp, p));
368 #endif
369 #if defined(GPROF) || defined(DDBPROF)
370 	case KERN_PROF:
371 		return (sysctl_doprof(name, namelen, oldp, oldlenp,
372 		    newp, newlen));
373 #endif
374 	case KERN_MALLOCSTATS:
375 		return (sysctl_malloc(name, namelen, oldp, oldlenp,
376 		    newp, newlen, p));
377 	case KERN_TTY:
378 		return (sysctl_tty(name, namelen, oldp, oldlenp,
379 		    newp, newlen));
380 	case KERN_POOL:
381 		return (sysctl_dopool(name, namelen, oldp, oldlenp));
382 #if defined(SYSVMSG) || defined(SYSVSEM) || defined(SYSVSHM)
383 	case KERN_SYSVIPC_INFO:
384 		return (sysctl_sysvipc(name, namelen, oldp, oldlenp));
385 #endif
386 #ifdef SYSVSEM
387 	case KERN_SEMINFO:
388 		return (sysctl_sysvsem(name, namelen, oldp, oldlenp,
389 		    newp, newlen));
390 #endif
391 #ifdef SYSVSHM
392 	case KERN_SHMINFO:
393 		return (sysctl_sysvshm(name, namelen, oldp, oldlenp,
394 		    newp, newlen));
395 #endif
396 #ifndef SMALL_KERNEL
397 	case KERN_INTRCNT:
398 		return (sysctl_intrcnt(name, namelen, oldp, oldlenp));
399 	case KERN_WATCHDOG:
400 		return (sysctl_wdog(name, namelen, oldp, oldlenp,
401 		    newp, newlen));
402 #endif
403 #ifndef SMALL_KERNEL
404 	case KERN_EVCOUNT:
405 		return (evcount_sysctl(name, namelen, oldp, oldlenp,
406 		    newp, newlen));
407 #endif
408 	case KERN_TIMECOUNTER:
409 		return (sysctl_tc(name, namelen, oldp, oldlenp, newp, newlen));
410 	case KERN_CPTIME2:
411 		return (sysctl_cptime2(name, namelen, oldp, oldlenp,
412 		    newp, newlen));
413 #ifdef WITNESS
414 	case KERN_WITNESSWATCH:
415 		return witness_sysctl_watch(oldp, oldlenp, newp, newlen);
416 	case KERN_WITNESS:
417 		return witness_sysctl(name, namelen, oldp, oldlenp,
418 		    newp, newlen);
419 #endif
420 #if NAUDIO > 0
421 	case KERN_AUDIO:
422 		return (sysctl_audio(name, namelen, oldp, oldlenp,
423 		    newp, newlen));
424 #endif
425 #if NVIDEO > 0
426 	case KERN_VIDEO:
427 		return (sysctl_video(name, namelen, oldp, oldlenp,
428 		    newp, newlen));
429 #endif
430 	case KERN_CPUSTATS:
431 		return (sysctl_cpustats(name, namelen, oldp, oldlenp,
432 		    newp, newlen));
433 	case KERN_CLOCKINTR:
434 		return sysctl_clockintr(name, namelen, oldp, oldlenp, newp,
435 		    newlen);
436 	default:
437 		return (ENOTDIR);	/* overloaded */
438 	}
439 }
440 
441 /*
442  * kernel related system variables.
443  */
444 int
445 kern_sysctl(int *name, u_int namelen, void *oldp, size_t *oldlenp, void *newp,
446     size_t newlen, struct proc *p)
447 {
448 	int error, level, inthostid, stackgap;
449 	dev_t dev;
450 	extern int pool_debug;
451 
452 	/* dispatch the non-terminal nodes first */
453 	if (namelen != 1) {
454 		return kern_sysctl_dirs(name[0], name + 1, namelen - 1,
455 		    oldp, oldlenp, newp, newlen, p);
456 	}
457 
458 	switch (name[0]) {
459 	case KERN_OSTYPE:
460 		return (sysctl_rdstring(oldp, oldlenp, newp, ostype));
461 	case KERN_OSRELEASE:
462 		return (sysctl_rdstring(oldp, oldlenp, newp, osrelease));
463 	case KERN_OSVERSION:
464 		return (sysctl_rdstring(oldp, oldlenp, newp, osversion));
465 	case KERN_VERSION:
466 		return (sysctl_rdstring(oldp, oldlenp, newp, version));
467 	case KERN_NUMVNODES:  /* XXX numvnodes is a long */
468 		return (sysctl_rdint(oldp, oldlenp, newp, numvnodes));
469 	case KERN_SECURELVL:
470 		level = securelevel;
471 		if ((error = sysctl_int(oldp, oldlenp, newp, newlen, &level)) ||
472 		    newp == NULL)
473 			return (error);
474 		if ((securelevel > 0 || level < -1) &&
475 		    level < securelevel && p->p_p->ps_pid != 1)
476 			return (EPERM);
477 		securelevel = level;
478 		return (0);
479 #if NDT > 0
480 	case KERN_ALLOWDT:
481 		return (sysctl_securelevel_int(oldp, oldlenp, newp, newlen,
482 		    &allowdt));
483 #endif
484 	case KERN_ALLOWKMEM:
485 		return (sysctl_securelevel_int(oldp, oldlenp, newp, newlen,
486 		    &allowkmem));
487 	case KERN_HOSTNAME:
488 		error = sysctl_tstring(oldp, oldlenp, newp, newlen,
489 		    hostname, sizeof(hostname));
490 		if (newp && !error)
491 			hostnamelen = newlen;
492 		return (error);
493 	case KERN_DOMAINNAME:
494 		if (securelevel >= 1 && domainnamelen && newp)
495 			error = EPERM;
496 		else
497 			error = sysctl_tstring(oldp, oldlenp, newp, newlen,
498 			    domainname, sizeof(domainname));
499 		if (newp && !error)
500 			domainnamelen = newlen;
501 		return (error);
502 	case KERN_HOSTID:
503 		inthostid = hostid;  /* XXX assumes sizeof long <= sizeof int */
504 		error =  sysctl_int(oldp, oldlenp, newp, newlen, &inthostid);
505 		hostid = inthostid;
506 		return (error);
507 	case KERN_CLOCKRATE:
508 		return (sysctl_clockrate(oldp, oldlenp, newp));
509 	case KERN_BOOTTIME: {
510 		struct timeval bt;
511 		memset(&bt, 0, sizeof bt);
512 		microboottime(&bt);
513 		return (sysctl_rdstruct(oldp, oldlenp, newp, &bt, sizeof bt));
514 	  }
515 	case KERN_MBSTAT: {
516 		extern struct cpumem *mbstat;
517 		uint64_t counters[MBSTAT_COUNT];
518 		struct mbstat mbs;
519 		unsigned int i;
520 
521 		memset(&mbs, 0, sizeof(mbs));
522 		counters_read(mbstat, counters, MBSTAT_COUNT, NULL);
523 		for (i = 0; i < MBSTAT_TYPES; i++)
524 			mbs.m_mtypes[i] = counters[i];
525 
526 		mbs.m_drops = counters[MBSTAT_DROPS];
527 		mbs.m_wait = counters[MBSTAT_WAIT];
528 		mbs.m_drain = counters[MBSTAT_DRAIN];
529 
530 		return (sysctl_rdstruct(oldp, oldlenp, newp,
531 		    &mbs, sizeof(mbs)));
532 	}
533 	case KERN_MSGBUFSIZE:
534 	case KERN_CONSBUFSIZE: {
535 		struct msgbuf *mp;
536 		mp = (name[0] == KERN_MSGBUFSIZE) ? msgbufp : consbufp;
537 		/*
538 		 * deal with cases where the message buffer has
539 		 * become corrupted.
540 		 */
541 		if (!mp || mp->msg_magic != MSG_MAGIC)
542 			return (ENXIO);
543 		return (sysctl_rdint(oldp, oldlenp, newp, mp->msg_bufs));
544 	}
545 	case KERN_CONSBUF:
546 		if ((error = suser(p)))
547 			return (error);
548 		/* FALLTHROUGH */
549 	case KERN_MSGBUF: {
550 		struct msgbuf *mp;
551 		mp = (name[0] == KERN_MSGBUF) ? msgbufp : consbufp;
552 		/* see note above */
553 		if (!mp || mp->msg_magic != MSG_MAGIC)
554 			return (ENXIO);
555 		return (sysctl_rdstruct(oldp, oldlenp, newp, mp,
556 		    mp->msg_bufs + offsetof(struct msgbuf, msg_bufc)));
557 	}
558 	case KERN_CPTIME:
559 	{
560 		CPU_INFO_ITERATOR cii;
561 		struct cpu_info *ci;
562 		long cp_time[CPUSTATES];
563 		int i, n = 0;
564 
565 		memset(cp_time, 0, sizeof(cp_time));
566 
567 		CPU_INFO_FOREACH(cii, ci) {
568 			if (!cpu_is_online(ci))
569 				continue;
570 			n++;
571 			for (i = 0; i < CPUSTATES; i++)
572 				cp_time[i] += ci->ci_schedstate.spc_cp_time[i];
573 		}
574 
575 		for (i = 0; i < CPUSTATES; i++)
576 			cp_time[i] /= n;
577 
578 		return (sysctl_rdstruct(oldp, oldlenp, newp, &cp_time,
579 		    sizeof(cp_time)));
580 	}
581 	case KERN_NCHSTATS:
582 		return (sysctl_rdstruct(oldp, oldlenp, newp, &nchstats,
583 		    sizeof(struct nchstats)));
584 	case KERN_FORKSTAT:
585 		return (sysctl_rdstruct(oldp, oldlenp, newp, &forkstat,
586 		    sizeof(struct forkstat)));
587 	case KERN_STACKGAPRANDOM:
588 		stackgap = stackgap_random;
589 		error = sysctl_int(oldp, oldlenp, newp, newlen, &stackgap);
590 		if (error)
591 			return (error);
592 		/*
593 		 * Safety harness.
594 		 */
595 		if ((stackgap < ALIGNBYTES && stackgap != 0) ||
596 		    !powerof2(stackgap) || stackgap >= MAXSSIZ)
597 			return (EINVAL);
598 		stackgap_random = stackgap;
599 		return (0);
600 	case KERN_MAXCLUSTERS: {
601 		int val = nmbclust;
602 		error = sysctl_int(oldp, oldlenp, newp, newlen, &val);
603 		if (error == 0 && val != nmbclust)
604 			error = nmbclust_update(val);
605 		return (error);
606 	}
607 	case KERN_CACHEPCT: {
608 		u_int64_t dmapages;
609 		int opct, pgs;
610 		opct = bufcachepercent;
611 		error = sysctl_int(oldp, oldlenp, newp, newlen,
612 		    &bufcachepercent);
613 		if (error)
614 			return(error);
615 		if (bufcachepercent > 90 || bufcachepercent < 5) {
616 			bufcachepercent = opct;
617 			return (EINVAL);
618 		}
619 		dmapages = uvm_pagecount(&dma_constraint);
620 		if (bufcachepercent != opct) {
621 			pgs = bufcachepercent * dmapages / 100;
622 			bufadjust(pgs); /* adjust bufpages */
623 			bufhighpages = bufpages; /* set high water mark */
624 		}
625 		return(0);
626 	}
627 	case KERN_CONSDEV:
628 		if (cn_tab != NULL)
629 			dev = cn_tab->cn_dev;
630 		else
631 			dev = NODEV;
632 		return sysctl_rdstruct(oldp, oldlenp, newp, &dev, sizeof(dev));
633 	case KERN_POOL_DEBUG: {
634 		int old_pool_debug = pool_debug;
635 
636 		error = sysctl_int(oldp, oldlenp, newp, newlen,
637 		    &pool_debug);
638 		if (error == 0 && pool_debug != old_pool_debug)
639 			pool_reclaim_all();
640 		return (error);
641 	}
642 #if NPF > 0
643 	case KERN_PFSTATUS:
644 		return (pf_sysctl(oldp, oldlenp, newp, newlen));
645 #endif
646 	case KERN_TIMEOUT_STATS:
647 		return (timeout_sysctl(oldp, oldlenp, newp, newlen));
648 	case KERN_UTC_OFFSET:
649 		return (sysctl_utc_offset(oldp, oldlenp, newp, newlen));
650 	default:
651 		return (sysctl_bounded_arr(kern_vars, nitems(kern_vars), name,
652 		    namelen, oldp, oldlenp, newp, newlen));
653 	}
654 	/* NOTREACHED */
655 }
656 
657 /*
658  * hardware related system variables.
659  */
660 char *hw_vendor, *hw_prod, *hw_uuid, *hw_serial, *hw_ver;
661 int allowpowerdown = 1;
662 int hw_power = 1;
663 
664 /* morally const values reported by sysctl_bounded_arr */
665 static int byte_order = BYTE_ORDER;
666 static int page_size = PAGE_SIZE;
667 
668 const struct sysctl_bounded_args hw_vars[] = {
669 	{HW_NCPU, &ncpus, SYSCTL_INT_READONLY},
670 	{HW_NCPUFOUND, &ncpusfound, SYSCTL_INT_READONLY},
671 	{HW_BYTEORDER, &byte_order, SYSCTL_INT_READONLY},
672 	{HW_PAGESIZE, &page_size, SYSCTL_INT_READONLY},
673 	{HW_DISKCOUNT, &disk_count, SYSCTL_INT_READONLY},
674 	{HW_POWER, &hw_power, SYSCTL_INT_READONLY},
675 };
676 
677 int
678 hw_sysctl(int *name, u_int namelen, void *oldp, size_t *oldlenp, void *newp,
679     size_t newlen, struct proc *p)
680 {
681 	extern char machine[], cpu_model[];
682 	int err, cpuspeed;
683 
684 	/*
685 	 * all sysctl names at this level except sensors and battery
686 	 * are terminal
687 	 */
688 	if (name[0] != HW_SENSORS && name[0] != HW_BATTERY && namelen != 1)
689 		return (ENOTDIR);		/* overloaded */
690 
691 	switch (name[0]) {
692 	case HW_MACHINE:
693 		return (sysctl_rdstring(oldp, oldlenp, newp, machine));
694 	case HW_MODEL:
695 		return (sysctl_rdstring(oldp, oldlenp, newp, cpu_model));
696 	case HW_NCPUONLINE:
697 		return (sysctl_rdint(oldp, oldlenp, newp,
698 		    sysctl_hwncpuonline()));
699 	case HW_PHYSMEM:
700 		return (sysctl_rdint(oldp, oldlenp, newp, ptoa(physmem)));
701 	case HW_USERMEM:
702 		return (sysctl_rdint(oldp, oldlenp, newp,
703 		    ptoa(physmem - uvmexp.wired)));
704 	case HW_DISKNAMES:
705 		err = sysctl_diskinit(0, p);
706 		if (err)
707 			return err;
708 		if (disknames)
709 			return (sysctl_rdstring(oldp, oldlenp, newp,
710 			    disknames));
711 		else
712 			return (sysctl_rdstring(oldp, oldlenp, newp, ""));
713 	case HW_DISKSTATS:
714 		err = sysctl_diskinit(1, p);
715 		if (err)
716 			return err;
717 		return (sysctl_rdstruct(oldp, oldlenp, newp, diskstats,
718 		    disk_count * sizeof(struct diskstats)));
719 	case HW_CPUSPEED:
720 		if (!cpu_cpuspeed)
721 			return (EOPNOTSUPP);
722 		err = cpu_cpuspeed(&cpuspeed);
723 		if (err)
724 			return err;
725 		return (sysctl_rdint(oldp, oldlenp, newp, cpuspeed));
726 #ifndef	SMALL_KERNEL
727 	case HW_SENSORS:
728 		return (sysctl_sensors(name + 1, namelen - 1, oldp, oldlenp,
729 		    newp, newlen));
730 	case HW_SETPERF:
731 		return (sysctl_hwsetperf(oldp, oldlenp, newp, newlen));
732 	case HW_PERFPOLICY:
733 		return (sysctl_hwperfpolicy(oldp, oldlenp, newp, newlen));
734 #endif /* !SMALL_KERNEL */
735 	case HW_VENDOR:
736 		if (hw_vendor)
737 			return (sysctl_rdstring(oldp, oldlenp, newp,
738 			    hw_vendor));
739 		else
740 			return (EOPNOTSUPP);
741 	case HW_PRODUCT:
742 		if (hw_prod)
743 			return (sysctl_rdstring(oldp, oldlenp, newp, hw_prod));
744 		else
745 			return (EOPNOTSUPP);
746 	case HW_VERSION:
747 		if (hw_ver)
748 			return (sysctl_rdstring(oldp, oldlenp, newp, hw_ver));
749 		else
750 			return (EOPNOTSUPP);
751 	case HW_SERIALNO:
752 		if (hw_serial)
753 			return (sysctl_rdstring(oldp, oldlenp, newp,
754 			    hw_serial));
755 		else
756 			return (EOPNOTSUPP);
757 	case HW_UUID:
758 		if (hw_uuid)
759 			return (sysctl_rdstring(oldp, oldlenp, newp, hw_uuid));
760 		else
761 			return (EOPNOTSUPP);
762 	case HW_PHYSMEM64:
763 		return (sysctl_rdquad(oldp, oldlenp, newp,
764 		    ptoa((psize_t)physmem)));
765 	case HW_USERMEM64:
766 		return (sysctl_rdquad(oldp, oldlenp, newp,
767 		    ptoa((psize_t)physmem - uvmexp.wired)));
768 	case HW_ALLOWPOWERDOWN:
769 		return (sysctl_securelevel_int(oldp, oldlenp, newp, newlen,
770 		    &allowpowerdown));
771 #ifdef __HAVE_CPU_TOPOLOGY
772 	case HW_SMT:
773 		return (sysctl_hwsmt(oldp, oldlenp, newp, newlen));
774 #endif
775 #ifndef SMALL_KERNEL
776 	case HW_BATTERY:
777 		return (sysctl_hwbattery(name + 1, namelen - 1, oldp, oldlenp,
778 		    newp, newlen));
779 #endif
780 	default:
781 		return sysctl_bounded_arr(hw_vars, nitems(hw_vars), name,
782 		    namelen, oldp, oldlenp, newp, newlen);
783 	}
784 	/* NOTREACHED */
785 }
786 
787 #ifndef SMALL_KERNEL
788 
789 int hw_battery_chargemode;
790 int hw_battery_chargestart;
791 int hw_battery_chargestop;
792 int (*hw_battery_setchargemode)(int);
793 int (*hw_battery_setchargestart)(int);
794 int (*hw_battery_setchargestop)(int);
795 
796 int
797 sysctl_hwchargemode(void *oldp, size_t *oldlenp, void *newp, size_t newlen)
798 {
799 	int mode = hw_battery_chargemode;
800 	int error;
801 
802 	if (!hw_battery_setchargemode)
803 		return EOPNOTSUPP;
804 
805 	error = sysctl_int_bounded(oldp, oldlenp, newp, newlen,
806 	    &mode, -1, 1);
807 	if (error)
808 		return error;
809 
810 	if (newp != NULL)
811 		error = hw_battery_setchargemode(mode);
812 
813 	return error;
814 }
815 
816 int
817 sysctl_hwchargestart(void *oldp, size_t *oldlenp, void *newp, size_t newlen)
818 {
819 	int start = hw_battery_chargestart;
820 	int error;
821 
822 	if (!hw_battery_setchargestart)
823 		return EOPNOTSUPP;
824 
825 	error = sysctl_int_bounded(oldp, oldlenp, newp, newlen,
826 	    &start, 0, 100);
827 	if (error)
828 		return error;
829 
830 	if (newp != NULL)
831 		error = hw_battery_setchargestart(start);
832 
833 	return error;
834 }
835 
836 int
837 sysctl_hwchargestop(void *oldp, size_t *oldlenp, void *newp, size_t newlen)
838 {
839 	int stop = hw_battery_chargestop;
840 	int error;
841 
842 	if (!hw_battery_setchargestop)
843 		return EOPNOTSUPP;
844 
845 	error = sysctl_int_bounded(oldp, oldlenp, newp, newlen,
846 	    &stop, 0, 100);
847 	if (error)
848 		return error;
849 
850 	if (newp != NULL)
851 		error = hw_battery_setchargestop(stop);
852 
853 	return error;
854 }
855 
856 int
857 sysctl_hwbattery(int *name, u_int namelen, void *oldp, size_t *oldlenp,
858     void *newp, size_t newlen)
859 {
860 	if (namelen != 1)
861 		return (ENOTDIR);
862 
863 	switch (name[0]) {
864 	case HW_BATTERY_CHARGEMODE:
865 		return (sysctl_hwchargemode(oldp, oldlenp, newp, newlen));
866 	case HW_BATTERY_CHARGESTART:
867 		return (sysctl_hwchargestart(oldp, oldlenp, newp, newlen));
868 	case HW_BATTERY_CHARGESTOP:
869 		return (sysctl_hwchargestop(oldp, oldlenp, newp, newlen));
870 	default:
871 		return (EOPNOTSUPP);
872 	}
873 	/* NOTREACHED */
874 }
875 
876 #endif
877 
878 #ifdef DEBUG_SYSCTL
879 /*
880  * Debugging related system variables.
881  */
882 extern struct ctldebug debug_vfs_busyprt;
883 struct ctldebug debug1, debug2, debug3, debug4;
884 struct ctldebug debug5, debug6, debug7, debug8, debug9;
885 struct ctldebug debug10, debug11, debug12, debug13, debug14;
886 struct ctldebug debug15, debug16, debug17, debug18, debug19;
887 static struct ctldebug *debugvars[CTL_DEBUG_MAXID] = {
888 	&debug_vfs_busyprt,
889 	&debug1, &debug2, &debug3, &debug4,
890 	&debug5, &debug6, &debug7, &debug8, &debug9,
891 	&debug10, &debug11, &debug12, &debug13, &debug14,
892 	&debug15, &debug16, &debug17, &debug18, &debug19,
893 };
894 int
895 debug_sysctl(int *name, u_int namelen, void *oldp, size_t *oldlenp, void *newp,
896     size_t newlen, struct proc *p)
897 {
898 	struct ctldebug *cdp;
899 
900 	/* all sysctl names at this level are name and field */
901 	if (namelen != 2)
902 		return (ENOTDIR);		/* overloaded */
903 	if (name[0] < 0 || name[0] >= nitems(debugvars))
904 		return (EOPNOTSUPP);
905 	cdp = debugvars[name[0]];
906 	if (cdp->debugname == 0)
907 		return (EOPNOTSUPP);
908 	switch (name[1]) {
909 	case CTL_DEBUG_NAME:
910 		return (sysctl_rdstring(oldp, oldlenp, newp, cdp->debugname));
911 	case CTL_DEBUG_VALUE:
912 		return (sysctl_int(oldp, oldlenp, newp, newlen, cdp->debugvar));
913 	default:
914 		return (EOPNOTSUPP);
915 	}
916 	/* NOTREACHED */
917 }
918 #endif /* DEBUG_SYSCTL */
919 
920 /*
921  * Reads, or writes that lower the value
922  */
923 int
924 sysctl_int_lower(void *oldp, size_t *oldlenp, void *newp, size_t newlen,
925     int *valp)
926 {
927 	unsigned int oval = *valp, val = *valp;
928 	int error;
929 
930 	if (newp == NULL)
931 		return (sysctl_rdint(oldp, oldlenp, newp, val));
932 
933 	if ((error = sysctl_int(oldp, oldlenp, newp, newlen, &val)))
934 		return (error);
935 	if (val > oval)
936 		return (EPERM);		/* do not allow raising */
937 	*(unsigned int *)valp = val;
938 	return (0);
939 }
940 
941 /*
942  * Validate parameters and get old / set new parameters
943  * for an integer-valued sysctl function.
944  */
945 int
946 sysctl_int(void *oldp, size_t *oldlenp, void *newp, size_t newlen, int *valp)
947 {
948 	int error = 0;
949 
950 	if (oldp && *oldlenp < sizeof(int))
951 		return (ENOMEM);
952 	if (newp && newlen != sizeof(int))
953 		return (EINVAL);
954 	*oldlenp = sizeof(int);
955 	if (oldp)
956 		error = copyout(valp, oldp, sizeof(int));
957 	if (error == 0 && newp)
958 		error = copyin(newp, valp, sizeof(int));
959 	return (error);
960 }
961 
962 /*
963  * As above, but read-only.
964  */
965 int
966 sysctl_rdint(void *oldp, size_t *oldlenp, void *newp, int val)
967 {
968 	int error = 0;
969 
970 	if (oldp && *oldlenp < sizeof(int))
971 		return (ENOMEM);
972 	if (newp)
973 		return (EPERM);
974 	*oldlenp = sizeof(int);
975 	if (oldp)
976 		error = copyout((caddr_t)&val, oldp, sizeof(int));
977 	return (error);
978 }
979 
980 /*
981  * Selects between sysctl_rdint and sysctl_int according to securelevel.
982  */
983 int
984 sysctl_securelevel_int(void *oldp, size_t *oldlenp, void *newp, size_t newlen,
985     int *valp)
986 {
987 	if (securelevel > 0)
988 		return (sysctl_rdint(oldp, oldlenp, newp, *valp));
989 	return (sysctl_int(oldp, oldlenp, newp, newlen, valp));
990 }
991 
992 /*
993  * Read-only or bounded integer values.
994  */
995 int
996 sysctl_int_bounded(void *oldp, size_t *oldlenp, void *newp, size_t newlen,
997     int *valp, int minimum, int maximum)
998 {
999 	int val = *valp;
1000 	int error;
1001 
1002 	/* read only */
1003 	if (newp == NULL || minimum > maximum)
1004 		return (sysctl_rdint(oldp, oldlenp, newp, val));
1005 
1006 	if ((error = sysctl_int(oldp, oldlenp, newp, newlen, &val)))
1007 		return (error);
1008 	/* outside limits */
1009 	if (val < minimum || maximum < val)
1010 		return (EINVAL);
1011 	*valp = val;
1012 	return (0);
1013 }
1014 
1015 /*
1016  * Array of read-only or bounded integer values.
1017  */
1018 int
1019 sysctl_bounded_arr(const struct sysctl_bounded_args *valpp, u_int valplen,
1020     int *name, u_int namelen, void *oldp, size_t *oldlenp, void *newp,
1021     size_t newlen)
1022 {
1023 	u_int i;
1024 	if (namelen != 1)
1025 		return (ENOTDIR);
1026 	for (i = 0; i < valplen; ++i) {
1027 		if (valpp[i].mib == name[0]) {
1028 			return (sysctl_int_bounded(oldp, oldlenp, newp, newlen,
1029 			    valpp[i].var, valpp[i].minimum, valpp[i].maximum));
1030 		}
1031 	}
1032 	return (EOPNOTSUPP);
1033 }
1034 
1035 /*
1036  * Validate parameters and get old / set new parameters
1037  * for an integer-valued sysctl function.
1038  */
1039 int
1040 sysctl_quad(void *oldp, size_t *oldlenp, void *newp, size_t newlen,
1041     int64_t *valp)
1042 {
1043 	int error = 0;
1044 
1045 	if (oldp && *oldlenp < sizeof(int64_t))
1046 		return (ENOMEM);
1047 	if (newp && newlen != sizeof(int64_t))
1048 		return (EINVAL);
1049 	*oldlenp = sizeof(int64_t);
1050 	if (oldp)
1051 		error = copyout(valp, oldp, sizeof(int64_t));
1052 	if (error == 0 && newp)
1053 		error = copyin(newp, valp, sizeof(int64_t));
1054 	return (error);
1055 }
1056 
1057 /*
1058  * As above, but read-only.
1059  */
1060 int
1061 sysctl_rdquad(void *oldp, size_t *oldlenp, void *newp, int64_t val)
1062 {
1063 	int error = 0;
1064 
1065 	if (oldp && *oldlenp < sizeof(int64_t))
1066 		return (ENOMEM);
1067 	if (newp)
1068 		return (EPERM);
1069 	*oldlenp = sizeof(int64_t);
1070 	if (oldp)
1071 		error = copyout((caddr_t)&val, oldp, sizeof(int64_t));
1072 	return (error);
1073 }
1074 
1075 /*
1076  * Validate parameters and get old / set new parameters
1077  * for a string-valued sysctl function.
1078  */
1079 int
1080 sysctl_string(void *oldp, size_t *oldlenp, void *newp, size_t newlen, char *str,
1081     size_t maxlen)
1082 {
1083 	return sysctl__string(oldp, oldlenp, newp, newlen, str, maxlen, 0);
1084 }
1085 
1086 int
1087 sysctl_tstring(void *oldp, size_t *oldlenp, void *newp, size_t newlen,
1088     char *str, size_t maxlen)
1089 {
1090 	return sysctl__string(oldp, oldlenp, newp, newlen, str, maxlen, 1);
1091 }
1092 
1093 int
1094 sysctl__string(void *oldp, size_t *oldlenp, void *newp, size_t newlen,
1095     char *str, size_t maxlen, int trunc)
1096 {
1097 	size_t len;
1098 	int error = 0;
1099 
1100 	len = strlen(str) + 1;
1101 	if (oldp && *oldlenp < len) {
1102 		if (trunc == 0 || *oldlenp == 0)
1103 			return (ENOMEM);
1104 	}
1105 	if (newp && newlen >= maxlen)
1106 		return (EINVAL);
1107 	if (oldp) {
1108 		if (trunc && *oldlenp < len) {
1109 			len = *oldlenp;
1110 			error = copyout(str, oldp, len - 1);
1111 			if (error == 0)
1112 				error = copyout("", (char *)oldp + len - 1, 1);
1113 		} else {
1114 			error = copyout(str, oldp, len);
1115 		}
1116 	}
1117 	*oldlenp = len;
1118 	if (error == 0 && newp) {
1119 		error = copyin(newp, str, newlen);
1120 		str[newlen] = 0;
1121 	}
1122 	return (error);
1123 }
1124 
1125 /*
1126  * As above, but read-only.
1127  */
1128 int
1129 sysctl_rdstring(void *oldp, size_t *oldlenp, void *newp, const char *str)
1130 {
1131 	size_t len;
1132 	int error = 0;
1133 
1134 	len = strlen(str) + 1;
1135 	if (oldp && *oldlenp < len)
1136 		return (ENOMEM);
1137 	if (newp)
1138 		return (EPERM);
1139 	*oldlenp = len;
1140 	if (oldp)
1141 		error = copyout(str, oldp, len);
1142 	return (error);
1143 }
1144 
1145 /*
1146  * Validate parameters and get old / set new parameters
1147  * for a structure oriented sysctl function.
1148  */
1149 int
1150 sysctl_struct(void *oldp, size_t *oldlenp, void *newp, size_t newlen, void *sp,
1151     size_t len)
1152 {
1153 	int error = 0;
1154 
1155 	if (oldp && *oldlenp < len)
1156 		return (ENOMEM);
1157 	if (newp && newlen > len)
1158 		return (EINVAL);
1159 	if (oldp) {
1160 		*oldlenp = len;
1161 		error = copyout(sp, oldp, len);
1162 	}
1163 	if (error == 0 && newp)
1164 		error = copyin(newp, sp, len);
1165 	return (error);
1166 }
1167 
1168 /*
1169  * Validate parameters and get old parameters
1170  * for a structure oriented sysctl function.
1171  */
1172 int
1173 sysctl_rdstruct(void *oldp, size_t *oldlenp, void *newp, const void *sp,
1174     size_t len)
1175 {
1176 	int error = 0;
1177 
1178 	if (oldp && *oldlenp < len)
1179 		return (ENOMEM);
1180 	if (newp)
1181 		return (EPERM);
1182 	*oldlenp = len;
1183 	if (oldp)
1184 		error = copyout(sp, oldp, len);
1185 	return (error);
1186 }
1187 
1188 #ifndef SMALL_KERNEL
1189 void
1190 fill_file(struct kinfo_file *kf, struct file *fp, struct filedesc *fdp,
1191 	  int fd, struct vnode *vp, struct process *pr, struct proc *p,
1192 	  struct socket *so, int show_pointers)
1193 {
1194 	struct vattr va;
1195 
1196 	memset(kf, 0, sizeof(*kf));
1197 
1198 	kf->fd_fd = fd;		/* might not really be an fd */
1199 
1200 	if (fp != NULL) {
1201 		if (show_pointers)
1202 			kf->f_fileaddr = PTRTOINT64(fp);
1203 		kf->f_flag = fp->f_flag;
1204 		kf->f_iflags = fp->f_iflags;
1205 		kf->f_type = fp->f_type;
1206 		kf->f_count = fp->f_count;
1207 		if (show_pointers)
1208 			kf->f_ucred = PTRTOINT64(fp->f_cred);
1209 		kf->f_uid = fp->f_cred->cr_uid;
1210 		kf->f_gid = fp->f_cred->cr_gid;
1211 		if (show_pointers)
1212 			kf->f_ops = PTRTOINT64(fp->f_ops);
1213 		if (show_pointers)
1214 			kf->f_data = PTRTOINT64(fp->f_data);
1215 		kf->f_usecount = 0;
1216 
1217 		if (suser(p) == 0 || p->p_ucred->cr_uid == fp->f_cred->cr_uid) {
1218 			mtx_enter(&fp->f_mtx);
1219 			kf->f_offset = fp->f_offset;
1220 			kf->f_rxfer = fp->f_rxfer;
1221 			kf->f_rwfer = fp->f_wxfer;
1222 			kf->f_seek = fp->f_seek;
1223 			kf->f_rbytes = fp->f_rbytes;
1224 			kf->f_wbytes = fp->f_wbytes;
1225 			mtx_leave(&fp->f_mtx);
1226 		} else
1227 			kf->f_offset = -1;
1228 	} else if (vp != NULL) {
1229 		/* fake it */
1230 		kf->f_type = DTYPE_VNODE;
1231 		kf->f_flag = FREAD;
1232 		if (fd == KERN_FILE_TRACE)
1233 			kf->f_flag |= FWRITE;
1234 	} else if (so != NULL) {
1235 		/* fake it */
1236 		kf->f_type = DTYPE_SOCKET;
1237 	}
1238 
1239 	/* information about the object associated with this file */
1240 	switch (kf->f_type) {
1241 	case DTYPE_VNODE:
1242 		if (fp != NULL)
1243 			vp = (struct vnode *)fp->f_data;
1244 
1245 		if (show_pointers)
1246 			kf->v_un = PTRTOINT64(vp->v_un.vu_socket);
1247 		kf->v_type = vp->v_type;
1248 		kf->v_tag = vp->v_tag;
1249 		kf->v_flag = vp->v_flag;
1250 		if (show_pointers)
1251 			kf->v_data = PTRTOINT64(vp->v_data);
1252 		if (show_pointers)
1253 			kf->v_mount = PTRTOINT64(vp->v_mount);
1254 		if (vp->v_mount)
1255 			strlcpy(kf->f_mntonname,
1256 			    vp->v_mount->mnt_stat.f_mntonname,
1257 			    sizeof(kf->f_mntonname));
1258 
1259 		if (VOP_GETATTR(vp, &va, p->p_ucred, p) == 0) {
1260 			kf->va_fileid = va.va_fileid;
1261 			kf->va_mode = MAKEIMODE(va.va_type, va.va_mode);
1262 			kf->va_size = va.va_size;
1263 			kf->va_rdev = va.va_rdev;
1264 			kf->va_fsid = va.va_fsid & 0xffffffff;
1265 			kf->va_nlink = va.va_nlink;
1266 		}
1267 		break;
1268 
1269 	case DTYPE_SOCKET: {
1270 		int locked = 0;
1271 
1272 		if (so == NULL) {
1273 			so = (struct socket *)fp->f_data;
1274 			/* if so is passed as parameter it is already locked */
1275 			switch (so->so_proto->pr_domain->dom_family) {
1276 			case AF_INET:
1277 			case AF_INET6:
1278 				NET_LOCK();
1279 				locked = 1;
1280 				break;
1281 			}
1282 		}
1283 
1284 		kf->so_type = so->so_type;
1285 		kf->so_state = so->so_state | so->so_snd.sb_state |
1286 		    so->so_rcv.sb_state;
1287 		if (show_pointers)
1288 			kf->so_pcb = PTRTOINT64(so->so_pcb);
1289 		else
1290 			kf->so_pcb = -1;
1291 		kf->so_protocol = so->so_proto->pr_protocol;
1292 		kf->so_family = so->so_proto->pr_domain->dom_family;
1293 		kf->so_rcv_cc = so->so_rcv.sb_cc;
1294 		kf->so_snd_cc = so->so_snd.sb_cc;
1295 		if (isspliced(so)) {
1296 			if (show_pointers)
1297 				kf->so_splice =
1298 				    PTRTOINT64(so->so_sp->ssp_socket);
1299 			kf->so_splicelen = so->so_sp->ssp_len;
1300 		} else if (issplicedback(so))
1301 			kf->so_splicelen = -1;
1302 		if (so->so_pcb == NULL) {
1303 			if (locked)
1304 				NET_UNLOCK();
1305 			break;
1306 		}
1307 		switch (kf->so_family) {
1308 		case AF_INET: {
1309 			struct inpcb *inpcb = so->so_pcb;
1310 
1311 			NET_ASSERT_LOCKED();
1312 			if (show_pointers)
1313 				kf->inp_ppcb = PTRTOINT64(inpcb->inp_ppcb);
1314 			kf->inp_lport = inpcb->inp_lport;
1315 			kf->inp_laddru[0] = inpcb->inp_laddr.s_addr;
1316 			kf->inp_fport = inpcb->inp_fport;
1317 			kf->inp_faddru[0] = inpcb->inp_faddr.s_addr;
1318 			kf->inp_rtableid = inpcb->inp_rtableid;
1319 			if (so->so_type == SOCK_RAW)
1320 				kf->inp_proto = inpcb->inp_ip.ip_p;
1321 			if (so->so_proto->pr_protocol == IPPROTO_TCP) {
1322 				struct tcpcb *tcpcb = (void *)inpcb->inp_ppcb;
1323 				kf->t_rcv_wnd = tcpcb->rcv_wnd;
1324 				kf->t_snd_wnd = tcpcb->snd_wnd;
1325 				kf->t_snd_cwnd = tcpcb->snd_cwnd;
1326 				kf->t_state = tcpcb->t_state;
1327 			}
1328 			break;
1329 		    }
1330 		case AF_INET6: {
1331 			struct inpcb *inpcb = so->so_pcb;
1332 
1333 			NET_ASSERT_LOCKED();
1334 			if (show_pointers)
1335 				kf->inp_ppcb = PTRTOINT64(inpcb->inp_ppcb);
1336 			kf->inp_lport = inpcb->inp_lport;
1337 			kf->inp_laddru[0] = inpcb->inp_laddr6.s6_addr32[0];
1338 			kf->inp_laddru[1] = inpcb->inp_laddr6.s6_addr32[1];
1339 			kf->inp_laddru[2] = inpcb->inp_laddr6.s6_addr32[2];
1340 			kf->inp_laddru[3] = inpcb->inp_laddr6.s6_addr32[3];
1341 			kf->inp_fport = inpcb->inp_fport;
1342 			kf->inp_faddru[0] = inpcb->inp_faddr6.s6_addr32[0];
1343 			kf->inp_faddru[1] = inpcb->inp_faddr6.s6_addr32[1];
1344 			kf->inp_faddru[2] = inpcb->inp_faddr6.s6_addr32[2];
1345 			kf->inp_faddru[3] = inpcb->inp_faddr6.s6_addr32[3];
1346 			kf->inp_rtableid = inpcb->inp_rtableid;
1347 			if (so->so_type == SOCK_RAW)
1348 				kf->inp_proto = inpcb->inp_ipv6.ip6_nxt;
1349 			if (so->so_proto->pr_protocol == IPPROTO_TCP) {
1350 				struct tcpcb *tcpcb = (void *)inpcb->inp_ppcb;
1351 				kf->t_rcv_wnd = tcpcb->rcv_wnd;
1352 				kf->t_snd_wnd = tcpcb->snd_wnd;
1353 				kf->t_state = tcpcb->t_state;
1354 			}
1355 			break;
1356 		    }
1357 		case AF_UNIX: {
1358 			struct unpcb *unpcb = so->so_pcb;
1359 
1360 			kf->f_msgcount = unpcb->unp_msgcount;
1361 			if (show_pointers) {
1362 				kf->unp_conn	= PTRTOINT64(unpcb->unp_conn);
1363 				kf->unp_refs	= PTRTOINT64(
1364 				    SLIST_FIRST(&unpcb->unp_refs));
1365 				kf->unp_nextref	= PTRTOINT64(
1366 				    SLIST_NEXT(unpcb, unp_nextref));
1367 				kf->v_un	= PTRTOINT64(unpcb->unp_vnode);
1368 				kf->unp_addr	= PTRTOINT64(unpcb->unp_addr);
1369 			}
1370 			if (unpcb->unp_addr != NULL) {
1371 				struct sockaddr_un *un = mtod(unpcb->unp_addr,
1372 				    struct sockaddr_un *);
1373 				memcpy(kf->unp_path, un->sun_path, un->sun_len
1374 				    - offsetof(struct sockaddr_un,sun_path));
1375 			}
1376 			break;
1377 		    }
1378 		}
1379 		if (locked)
1380 			NET_UNLOCK();
1381 		break;
1382 	    }
1383 
1384 	case DTYPE_PIPE: {
1385 		struct pipe *pipe = (struct pipe *)fp->f_data;
1386 
1387 		if (show_pointers)
1388 			kf->pipe_peer = PTRTOINT64(pipe->pipe_peer);
1389 		kf->pipe_state = pipe->pipe_state;
1390 		break;
1391 	    }
1392 
1393 	case DTYPE_KQUEUE: {
1394 		struct kqueue *kqi = (struct kqueue *)fp->f_data;
1395 
1396 		kf->kq_count = kqi->kq_count;
1397 		kf->kq_state = kqi->kq_state;
1398 		break;
1399 	    }
1400 	}
1401 
1402 	/* per-process information for KERN_FILE_BY[PU]ID */
1403 	if (pr != NULL) {
1404 		kf->p_pid = pr->ps_pid;
1405 		kf->p_uid = pr->ps_ucred->cr_uid;
1406 		kf->p_gid = pr->ps_ucred->cr_gid;
1407 		kf->p_tid = -1;
1408 		strlcpy(kf->p_comm, pr->ps_comm, sizeof(kf->p_comm));
1409 	}
1410 	if (fdp != NULL) {
1411 		fdplock(fdp);
1412 		kf->fd_ofileflags = fdp->fd_ofileflags[fd];
1413 		fdpunlock(fdp);
1414 	}
1415 }
1416 
1417 /*
1418  * Get file structures.
1419  */
1420 int
1421 sysctl_file(int *name, u_int namelen, char *where, size_t *sizep,
1422     struct proc *p)
1423 {
1424 	struct kinfo_file *kf;
1425 	struct filedesc *fdp;
1426 	struct file *fp;
1427 	struct process *pr;
1428 	size_t buflen, elem_size, elem_count, outsize;
1429 	char *dp = where;
1430 	int arg, i, error = 0, needed = 0, matched;
1431 	u_int op;
1432 	int show_pointers;
1433 
1434 	if (namelen > 4)
1435 		return (ENOTDIR);
1436 	if (namelen < 4 || name[2] > sizeof(*kf))
1437 		return (EINVAL);
1438 
1439 	buflen = where != NULL ? *sizep : 0;
1440 	op = name[0];
1441 	arg = name[1];
1442 	elem_size = name[2];
1443 	elem_count = name[3];
1444 	outsize = MIN(sizeof(*kf), elem_size);
1445 
1446 	if (elem_size < 1)
1447 		return (EINVAL);
1448 
1449 	show_pointers = suser(curproc) == 0;
1450 
1451 	kf = malloc(sizeof(*kf), M_TEMP, M_WAITOK);
1452 
1453 #define FILLIT2(fp, fdp, i, vp, pr, so) do {				\
1454 	if (buflen >= elem_size && elem_count > 0) {			\
1455 		fill_file(kf, fp, fdp, i, vp, pr, p, so, show_pointers);\
1456 		error = copyout(kf, dp, outsize);			\
1457 		if (error)						\
1458 			break;						\
1459 		dp += elem_size;					\
1460 		buflen -= elem_size;					\
1461 		elem_count--;						\
1462 	}								\
1463 	needed += elem_size;						\
1464 } while (0)
1465 #define FILLIT(fp, fdp, i, vp, pr) \
1466 	FILLIT2(fp, fdp, i, vp, pr, NULL)
1467 #define FILLSO(so) \
1468 	FILLIT2(NULL, NULL, 0, NULL, NULL, so)
1469 
1470 	switch (op) {
1471 	case KERN_FILE_BYFILE:
1472 		/* use the inp-tables to pick up closed connections, too */
1473 		if (arg == DTYPE_SOCKET) {
1474 			struct inpcb *inp;
1475 
1476 			NET_LOCK();
1477 			mtx_enter(&tcbtable.inpt_mtx);
1478 			TAILQ_FOREACH(inp, &tcbtable.inpt_queue, inp_queue)
1479 				FILLSO(inp->inp_socket);
1480 			mtx_leave(&tcbtable.inpt_mtx);
1481 			mtx_enter(&udbtable.inpt_mtx);
1482 			TAILQ_FOREACH(inp, &udbtable.inpt_queue, inp_queue)
1483 				FILLSO(inp->inp_socket);
1484 			mtx_leave(&udbtable.inpt_mtx);
1485 			mtx_enter(&rawcbtable.inpt_mtx);
1486 			TAILQ_FOREACH(inp, &rawcbtable.inpt_queue, inp_queue)
1487 				FILLSO(inp->inp_socket);
1488 			mtx_leave(&rawcbtable.inpt_mtx);
1489 #ifdef INET6
1490 			mtx_enter(&rawin6pcbtable.inpt_mtx);
1491 			TAILQ_FOREACH(inp, &rawin6pcbtable.inpt_queue,
1492 			    inp_queue)
1493 				FILLSO(inp->inp_socket);
1494 			mtx_leave(&rawin6pcbtable.inpt_mtx);
1495 #endif
1496 			NET_UNLOCK();
1497 		}
1498 		fp = NULL;
1499 		while ((fp = fd_iterfile(fp, p)) != NULL) {
1500 			if ((arg == 0 || fp->f_type == arg)) {
1501 				int af, skip = 0;
1502 				if (arg == DTYPE_SOCKET && fp->f_type == arg) {
1503 					af = ((struct socket *)fp->f_data)->
1504 					    so_proto->pr_domain->dom_family;
1505 					if (af == AF_INET || af == AF_INET6)
1506 						skip = 1;
1507 				}
1508 				if (!skip)
1509 					FILLIT(fp, NULL, 0, NULL, NULL);
1510 			}
1511 		}
1512 		break;
1513 	case KERN_FILE_BYPID:
1514 		/* A arg of -1 indicates all processes */
1515 		if (arg < -1) {
1516 			error = EINVAL;
1517 			break;
1518 		}
1519 		matched = 0;
1520 		LIST_FOREACH(pr, &allprocess, ps_list) {
1521 			/*
1522 			 * skip system, exiting, embryonic and undead
1523 			 * processes
1524 			 */
1525 			if (pr->ps_flags & (PS_SYSTEM | PS_EMBRYO | PS_EXITING))
1526 				continue;
1527 			if (arg > 0 && pr->ps_pid != (pid_t)arg) {
1528 				/* not the pid we are looking for */
1529 				continue;
1530 			}
1531 			matched = 1;
1532 			fdp = pr->ps_fd;
1533 			if (pr->ps_textvp)
1534 				FILLIT(NULL, NULL, KERN_FILE_TEXT, pr->ps_textvp, pr);
1535 			if (fdp->fd_cdir)
1536 				FILLIT(NULL, NULL, KERN_FILE_CDIR, fdp->fd_cdir, pr);
1537 			if (fdp->fd_rdir)
1538 				FILLIT(NULL, NULL, KERN_FILE_RDIR, fdp->fd_rdir, pr);
1539 			if (pr->ps_tracevp)
1540 				FILLIT(NULL, NULL, KERN_FILE_TRACE, pr->ps_tracevp, pr);
1541 			for (i = 0; i < fdp->fd_nfiles; i++) {
1542 				if ((fp = fd_getfile(fdp, i)) == NULL)
1543 					continue;
1544 				FILLIT(fp, fdp, i, NULL, pr);
1545 				FRELE(fp, p);
1546 			}
1547 		}
1548 		if (!matched)
1549 			error = ESRCH;
1550 		break;
1551 	case KERN_FILE_BYUID:
1552 		LIST_FOREACH(pr, &allprocess, ps_list) {
1553 			/*
1554 			 * skip system, exiting, embryonic and undead
1555 			 * processes
1556 			 */
1557 			if (pr->ps_flags & (PS_SYSTEM | PS_EMBRYO | PS_EXITING))
1558 				continue;
1559 			if (arg >= 0 && pr->ps_ucred->cr_uid != (uid_t)arg) {
1560 				/* not the uid we are looking for */
1561 				continue;
1562 			}
1563 			fdp = pr->ps_fd;
1564 			if (fdp->fd_cdir)
1565 				FILLIT(NULL, NULL, KERN_FILE_CDIR, fdp->fd_cdir, pr);
1566 			if (fdp->fd_rdir)
1567 				FILLIT(NULL, NULL, KERN_FILE_RDIR, fdp->fd_rdir, pr);
1568 			if (pr->ps_tracevp)
1569 				FILLIT(NULL, NULL, KERN_FILE_TRACE, pr->ps_tracevp, pr);
1570 			for (i = 0; i < fdp->fd_nfiles; i++) {
1571 				if ((fp = fd_getfile(fdp, i)) == NULL)
1572 					continue;
1573 				FILLIT(fp, fdp, i, NULL, pr);
1574 				FRELE(fp, p);
1575 			}
1576 		}
1577 		break;
1578 	default:
1579 		error = EINVAL;
1580 		break;
1581 	}
1582 	free(kf, M_TEMP, sizeof(*kf));
1583 
1584 	if (!error) {
1585 		if (where == NULL)
1586 			needed += KERN_FILESLOP * elem_size;
1587 		else if (*sizep < needed)
1588 			error = ENOMEM;
1589 		*sizep = needed;
1590 	}
1591 
1592 	return (error);
1593 }
1594 
1595 /*
1596  * try over estimating by 5 procs
1597  */
1598 #define KERN_PROCSLOP	5
1599 
1600 int
1601 sysctl_doproc(int *name, u_int namelen, char *where, size_t *sizep)
1602 {
1603 	struct kinfo_proc *kproc = NULL;
1604 	struct proc *p;
1605 	struct process *pr;
1606 	char *dp;
1607 	int arg, buflen, doingzomb, elem_size, elem_count;
1608 	int error, needed, op;
1609 	int dothreads = 0;
1610 	int show_pointers;
1611 
1612 	dp = where;
1613 	buflen = where != NULL ? *sizep : 0;
1614 	needed = error = 0;
1615 
1616 	if (namelen != 4 || name[2] <= 0 || name[3] < 0 ||
1617 	    name[2] > sizeof(*kproc))
1618 		return (EINVAL);
1619 	op = name[0];
1620 	arg = name[1];
1621 	elem_size = name[2];
1622 	elem_count = name[3];
1623 
1624 	dothreads = op & KERN_PROC_SHOW_THREADS;
1625 	op &= ~KERN_PROC_SHOW_THREADS;
1626 
1627 	show_pointers = suser(curproc) == 0;
1628 
1629 	if (where != NULL)
1630 		kproc = malloc(sizeof(*kproc), M_TEMP, M_WAITOK);
1631 
1632 	pr = LIST_FIRST(&allprocess);
1633 	doingzomb = 0;
1634 again:
1635 	for (; pr != NULL; pr = LIST_NEXT(pr, ps_list)) {
1636 		/* XXX skip processes in the middle of being zapped */
1637 		if (pr->ps_pgrp == NULL)
1638 			continue;
1639 
1640 		/*
1641 		 * Skip embryonic processes.
1642 		 */
1643 		if (pr->ps_flags & PS_EMBRYO)
1644 			continue;
1645 
1646 		/*
1647 		 * TODO - make more efficient (see notes below).
1648 		 */
1649 		switch (op) {
1650 
1651 		case KERN_PROC_PID:
1652 			/* could do this with just a lookup */
1653 			if (pr->ps_pid != (pid_t)arg)
1654 				continue;
1655 			break;
1656 
1657 		case KERN_PROC_PGRP:
1658 			/* could do this by traversing pgrp */
1659 			if (pr->ps_pgrp->pg_id != (pid_t)arg)
1660 				continue;
1661 			break;
1662 
1663 		case KERN_PROC_SESSION:
1664 			if (pr->ps_session->s_leader == NULL ||
1665 			    pr->ps_session->s_leader->ps_pid != (pid_t)arg)
1666 				continue;
1667 			break;
1668 
1669 		case KERN_PROC_TTY:
1670 			if ((pr->ps_flags & PS_CONTROLT) == 0 ||
1671 			    pr->ps_session->s_ttyp == NULL ||
1672 			    pr->ps_session->s_ttyp->t_dev != (dev_t)arg)
1673 				continue;
1674 			break;
1675 
1676 		case KERN_PROC_UID:
1677 			if (pr->ps_ucred->cr_uid != (uid_t)arg)
1678 				continue;
1679 			break;
1680 
1681 		case KERN_PROC_RUID:
1682 			if (pr->ps_ucred->cr_ruid != (uid_t)arg)
1683 				continue;
1684 			break;
1685 
1686 		case KERN_PROC_ALL:
1687 			if (pr->ps_flags & PS_SYSTEM)
1688 				continue;
1689 			break;
1690 
1691 		case KERN_PROC_KTHREAD:
1692 			/* no filtering */
1693 			break;
1694 
1695 		default:
1696 			error = EINVAL;
1697 			goto err;
1698 		}
1699 
1700 		if (buflen >= elem_size && elem_count > 0) {
1701 			fill_kproc(pr, kproc, NULL, show_pointers);
1702 			error = copyout(kproc, dp, elem_size);
1703 			if (error)
1704 				goto err;
1705 			dp += elem_size;
1706 			buflen -= elem_size;
1707 			elem_count--;
1708 		}
1709 		needed += elem_size;
1710 
1711 		/* Skip per-thread entries if not required by op */
1712 		if (!dothreads)
1713 			continue;
1714 
1715 		TAILQ_FOREACH(p, &pr->ps_threads, p_thr_link) {
1716 			if (buflen >= elem_size && elem_count > 0) {
1717 				fill_kproc(pr, kproc, p, show_pointers);
1718 				error = copyout(kproc, dp, elem_size);
1719 				if (error)
1720 					goto err;
1721 				dp += elem_size;
1722 				buflen -= elem_size;
1723 				elem_count--;
1724 			}
1725 			needed += elem_size;
1726 		}
1727 	}
1728 	if (doingzomb == 0) {
1729 		pr = LIST_FIRST(&zombprocess);
1730 		doingzomb++;
1731 		goto again;
1732 	}
1733 	if (where != NULL) {
1734 		*sizep = dp - where;
1735 		if (needed > *sizep) {
1736 			error = ENOMEM;
1737 			goto err;
1738 		}
1739 	} else {
1740 		needed += KERN_PROCSLOP * elem_size;
1741 		*sizep = needed;
1742 	}
1743 err:
1744 	if (kproc)
1745 		free(kproc, M_TEMP, sizeof(*kproc));
1746 	return (error);
1747 }
1748 
1749 /*
1750  * Fill in a kproc structure for the specified process.
1751  */
1752 void
1753 fill_kproc(struct process *pr, struct kinfo_proc *ki, struct proc *p,
1754     int show_pointers)
1755 {
1756 	struct session *s = pr->ps_session;
1757 	struct tty *tp;
1758 	struct vmspace *vm = pr->ps_vmspace;
1759 	struct timespec booted, st, ut, utc;
1760 	int isthread;
1761 
1762 	isthread = p != NULL;
1763 	if (!isthread)
1764 		p = pr->ps_mainproc;		/* XXX */
1765 
1766 	FILL_KPROC(ki, strlcpy, p, pr, pr->ps_ucred, pr->ps_pgrp,
1767 	    p, pr, s, vm, pr->ps_limit, pr->ps_sigacts, isthread,
1768 	    show_pointers);
1769 
1770 	/* stuff that's too painful to generalize into the macros */
1771 	if (pr->ps_pptr)
1772 		ki->p_ppid = pr->ps_ppid;
1773 	if (s->s_leader)
1774 		ki->p_sid = s->s_leader->ps_pid;
1775 
1776 	if ((pr->ps_flags & PS_CONTROLT) && (tp = s->s_ttyp)) {
1777 		ki->p_tdev = tp->t_dev;
1778 		ki->p_tpgid = tp->t_pgrp ? tp->t_pgrp->pg_id : -1;
1779 		if (show_pointers)
1780 			ki->p_tsess = PTRTOINT64(tp->t_session);
1781 	} else {
1782 		ki->p_tdev = NODEV;
1783 		ki->p_tpgid = -1;
1784 	}
1785 
1786 	/* fixups that can only be done in the kernel */
1787 	if ((pr->ps_flags & PS_ZOMBIE) == 0) {
1788 		if ((pr->ps_flags & PS_EMBRYO) == 0 && vm != NULL)
1789 			ki->p_vm_rssize = vm_resident_count(vm);
1790 		calctsru(isthread ? &p->p_tu : &pr->ps_tu, &ut, &st, NULL);
1791 		ki->p_uutime_sec = ut.tv_sec;
1792 		ki->p_uutime_usec = ut.tv_nsec/1000;
1793 		ki->p_ustime_sec = st.tv_sec;
1794 		ki->p_ustime_usec = st.tv_nsec/1000;
1795 
1796 		/* Convert starting uptime to a starting UTC time. */
1797 		nanoboottime(&booted);
1798 		timespecadd(&booted, &pr->ps_start, &utc);
1799 		ki->p_ustart_sec = utc.tv_sec;
1800 		ki->p_ustart_usec = utc.tv_nsec / 1000;
1801 
1802 #ifdef MULTIPROCESSOR
1803 		if (p->p_cpu != NULL)
1804 			ki->p_cpuid = CPU_INFO_UNIT(p->p_cpu);
1805 #endif
1806 	}
1807 
1808 	/* get %cpu and schedule state: just one thread or sum of all? */
1809 	if (isthread) {
1810 		ki->p_pctcpu = p->p_pctcpu;
1811 		ki->p_stat   = p->p_stat;
1812 	} else {
1813 		ki->p_pctcpu = 0;
1814 		ki->p_stat = (pr->ps_flags & PS_ZOMBIE) ? SDEAD : SIDL;
1815 		TAILQ_FOREACH(p, &pr->ps_threads, p_thr_link) {
1816 			ki->p_pctcpu += p->p_pctcpu;
1817 			/* find best state: ONPROC > RUN > STOP > SLEEP > .. */
1818 			if (p->p_stat == SONPROC || ki->p_stat == SONPROC)
1819 				ki->p_stat = SONPROC;
1820 			else if (p->p_stat == SRUN || ki->p_stat == SRUN)
1821 				ki->p_stat = SRUN;
1822 			else if (p->p_stat == SSTOP || ki->p_stat == SSTOP)
1823 				ki->p_stat = SSTOP;
1824 			else if (p->p_stat == SSLEEP)
1825 				ki->p_stat = SSLEEP;
1826 		}
1827 	}
1828 }
1829 
1830 int
1831 sysctl_proc_args(int *name, u_int namelen, void *oldp, size_t *oldlenp,
1832     struct proc *cp)
1833 {
1834 	struct process *vpr;
1835 	pid_t pid;
1836 	struct ps_strings pss;
1837 	struct iovec iov;
1838 	struct uio uio;
1839 	int error, cnt, op;
1840 	size_t limit;
1841 	char **rargv, **vargv;		/* reader vs. victim */
1842 	char *rarg, *varg, *buf;
1843 	struct vmspace *vm;
1844 	vaddr_t ps_strings;
1845 
1846 	if (namelen > 2)
1847 		return (ENOTDIR);
1848 	if (namelen < 2)
1849 		return (EINVAL);
1850 
1851 	pid = name[0];
1852 	op = name[1];
1853 
1854 	switch (op) {
1855 	case KERN_PROC_ARGV:
1856 	case KERN_PROC_NARGV:
1857 	case KERN_PROC_ENV:
1858 	case KERN_PROC_NENV:
1859 		break;
1860 	default:
1861 		return (EOPNOTSUPP);
1862 	}
1863 
1864 	if ((vpr = prfind(pid)) == NULL)
1865 		return (ESRCH);
1866 
1867 	if (oldp == NULL) {
1868 		if (op == KERN_PROC_NARGV || op == KERN_PROC_NENV)
1869 			*oldlenp = sizeof(int);
1870 		else
1871 			*oldlenp = ARG_MAX;	/* XXX XXX XXX */
1872 		return (0);
1873 	}
1874 
1875 	/* Either system process or exiting/zombie */
1876 	if (vpr->ps_flags & (PS_SYSTEM | PS_EXITING))
1877 		return (EINVAL);
1878 
1879 	/* Execing - danger. */
1880 	if ((vpr->ps_flags & PS_INEXEC))
1881 		return (EBUSY);
1882 
1883 	/* Only owner or root can get env */
1884 	if ((op == KERN_PROC_NENV || op == KERN_PROC_ENV) &&
1885 	    (vpr->ps_ucred->cr_uid != cp->p_ucred->cr_uid &&
1886 	    (error = suser(cp)) != 0))
1887 		return (error);
1888 
1889 	ps_strings = vpr->ps_strings;
1890 	vm = vpr->ps_vmspace;
1891 	uvmspace_addref(vm);
1892 	vpr = NULL;
1893 
1894 	buf = malloc(PAGE_SIZE, M_TEMP, M_WAITOK);
1895 
1896 	iov.iov_base = &pss;
1897 	iov.iov_len = sizeof(pss);
1898 	uio.uio_iov = &iov;
1899 	uio.uio_iovcnt = 1;
1900 	uio.uio_offset = (off_t)ps_strings;
1901 	uio.uio_resid = sizeof(pss);
1902 	uio.uio_segflg = UIO_SYSSPACE;
1903 	uio.uio_rw = UIO_READ;
1904 	uio.uio_procp = cp;
1905 
1906 	if ((error = uvm_io(&vm->vm_map, &uio, 0)) != 0)
1907 		goto out;
1908 
1909 	if (op == KERN_PROC_NARGV) {
1910 		error = sysctl_rdint(oldp, oldlenp, NULL, pss.ps_nargvstr);
1911 		goto out;
1912 	}
1913 	if (op == KERN_PROC_NENV) {
1914 		error = sysctl_rdint(oldp, oldlenp, NULL, pss.ps_nenvstr);
1915 		goto out;
1916 	}
1917 
1918 	if (op == KERN_PROC_ARGV) {
1919 		cnt = pss.ps_nargvstr;
1920 		vargv = pss.ps_argvstr;
1921 	} else {
1922 		cnt = pss.ps_nenvstr;
1923 		vargv = pss.ps_envstr;
1924 	}
1925 
1926 	/* -1 to have space for a terminating NUL */
1927 	limit = *oldlenp - 1;
1928 	*oldlenp = 0;
1929 
1930 	rargv = oldp;
1931 
1932 	/*
1933 	 * *oldlenp - number of bytes copied out into readers buffer.
1934 	 * limit - maximal number of bytes allowed into readers buffer.
1935 	 * rarg - pointer into readers buffer where next arg will be stored.
1936 	 * rargv - pointer into readers buffer where the next rarg pointer
1937 	 *  will be stored.
1938 	 * vargv - pointer into victim address space where the next argument
1939 	 *  will be read.
1940 	 */
1941 
1942 	/* space for cnt pointers and a NULL */
1943 	rarg = (char *)(rargv + cnt + 1);
1944 	*oldlenp += (cnt + 1) * sizeof(char **);
1945 
1946 	while (cnt > 0 && *oldlenp < limit) {
1947 		size_t len, vstrlen;
1948 
1949 		/* Write to readers argv */
1950 		if ((error = copyout(&rarg, rargv, sizeof(rarg))) != 0)
1951 			goto out;
1952 
1953 		/* read the victim argv */
1954 		iov.iov_base = &varg;
1955 		iov.iov_len = sizeof(varg);
1956 		uio.uio_iov = &iov;
1957 		uio.uio_iovcnt = 1;
1958 		uio.uio_offset = (off_t)(vaddr_t)vargv;
1959 		uio.uio_resid = sizeof(varg);
1960 		uio.uio_segflg = UIO_SYSSPACE;
1961 		uio.uio_rw = UIO_READ;
1962 		uio.uio_procp = cp;
1963 		if ((error = uvm_io(&vm->vm_map, &uio, 0)) != 0)
1964 			goto out;
1965 
1966 		if (varg == NULL)
1967 			break;
1968 
1969 		/*
1970 		 * read the victim arg. We must jump through hoops to avoid
1971 		 * crossing a page boundary too much and returning an error.
1972 		 */
1973 more:
1974 		len = PAGE_SIZE - (((vaddr_t)varg) & PAGE_MASK);
1975 		/* leave space for the terminating NUL */
1976 		iov.iov_base = buf;
1977 		iov.iov_len = len;
1978 		uio.uio_iov = &iov;
1979 		uio.uio_iovcnt = 1;
1980 		uio.uio_offset = (off_t)(vaddr_t)varg;
1981 		uio.uio_resid = len;
1982 		uio.uio_segflg = UIO_SYSSPACE;
1983 		uio.uio_rw = UIO_READ;
1984 		uio.uio_procp = cp;
1985 		if ((error = uvm_io(&vm->vm_map, &uio, 0)) != 0)
1986 			goto out;
1987 
1988 		for (vstrlen = 0; vstrlen < len; vstrlen++) {
1989 			if (buf[vstrlen] == '\0')
1990 				break;
1991 		}
1992 
1993 		/* Don't overflow readers buffer. */
1994 		if (*oldlenp + vstrlen + 1 >= limit) {
1995 			error = ENOMEM;
1996 			goto out;
1997 		}
1998 
1999 		if ((error = copyout(buf, rarg, vstrlen)) != 0)
2000 			goto out;
2001 
2002 		*oldlenp += vstrlen;
2003 		rarg += vstrlen;
2004 
2005 		/* The string didn't end in this page? */
2006 		if (vstrlen == len) {
2007 			varg += vstrlen;
2008 			goto more;
2009 		}
2010 
2011 		/* End of string. Terminate it with a NUL */
2012 		buf[0] = '\0';
2013 		if ((error = copyout(buf, rarg, 1)) != 0)
2014 			goto out;
2015 		*oldlenp += 1;
2016 		rarg += 1;
2017 
2018 		vargv++;
2019 		rargv++;
2020 		cnt--;
2021 	}
2022 
2023 	if (*oldlenp >= limit) {
2024 		error = ENOMEM;
2025 		goto out;
2026 	}
2027 
2028 	/* Write the terminating null */
2029 	rarg = NULL;
2030 	error = copyout(&rarg, rargv, sizeof(rarg));
2031 
2032 out:
2033 	uvmspace_free(vm);
2034 	free(buf, M_TEMP, PAGE_SIZE);
2035 	return (error);
2036 }
2037 
2038 int
2039 sysctl_proc_cwd(int *name, u_int namelen, void *oldp, size_t *oldlenp,
2040     struct proc *cp)
2041 {
2042 	struct process *findpr;
2043 	struct vnode *vp;
2044 	pid_t pid;
2045 	int error;
2046 	size_t lenused, len;
2047 	char *path, *bp, *bend;
2048 
2049 	if (namelen > 1)
2050 		return (ENOTDIR);
2051 	if (namelen < 1)
2052 		return (EINVAL);
2053 
2054 	pid = name[0];
2055 	if ((findpr = prfind(pid)) == NULL)
2056 		return (ESRCH);
2057 
2058 	if (oldp == NULL) {
2059 		*oldlenp = MAXPATHLEN * 4;
2060 		return (0);
2061 	}
2062 
2063 	/* Either system process or exiting/zombie */
2064 	if (findpr->ps_flags & (PS_SYSTEM | PS_EXITING))
2065 		return (EINVAL);
2066 
2067 	/* Only owner or root can get cwd */
2068 	if (findpr->ps_ucred->cr_uid != cp->p_ucred->cr_uid &&
2069 	    (error = suser(cp)) != 0)
2070 		return (error);
2071 
2072 	len = *oldlenp;
2073 	if (len > MAXPATHLEN * 4)
2074 		len = MAXPATHLEN * 4;
2075 	else if (len < 2)
2076 		return (ERANGE);
2077 	*oldlenp = 0;
2078 
2079 	/* snag a reference to the vnode before we can sleep */
2080 	vp = findpr->ps_fd->fd_cdir;
2081 	vref(vp);
2082 
2083 	path = malloc(len, M_TEMP, M_WAITOK);
2084 
2085 	bp = &path[len];
2086 	bend = bp;
2087 	*(--bp) = '\0';
2088 
2089 	/* Same as sys__getcwd */
2090 	error = vfs_getcwd_common(vp, NULL,
2091 	    &bp, path, len / 2, GETCWD_CHECK_ACCESS, cp);
2092 	if (error == 0) {
2093 		*oldlenp = lenused = bend - bp;
2094 		error = copyout(bp, oldp, lenused);
2095 	}
2096 
2097 	vrele(vp);
2098 	free(path, M_TEMP, len);
2099 
2100 	return (error);
2101 }
2102 
2103 int
2104 sysctl_proc_nobroadcastkill(int *name, u_int namelen, void *newp, size_t newlen,
2105     void *oldp, size_t *oldlenp, struct proc *cp)
2106 {
2107 	struct process *findpr;
2108 	pid_t pid;
2109 	int error, flag;
2110 
2111 	if (namelen > 1)
2112 		return (ENOTDIR);
2113 	if (namelen < 1)
2114 		return (EINVAL);
2115 
2116 	pid = name[0];
2117 	if ((findpr = prfind(pid)) == NULL)
2118 		return (ESRCH);
2119 
2120 	/* Either system process or exiting/zombie */
2121 	if (findpr->ps_flags & (PS_SYSTEM | PS_EXITING))
2122 		return (EINVAL);
2123 
2124 	/* Only root can change PS_NOBROADCASTKILL */
2125 	if (newp != NULL && (error = suser(cp)) != 0)
2126 		return (error);
2127 
2128 	/* get the PS_NOBROADCASTKILL flag */
2129 	flag = findpr->ps_flags & PS_NOBROADCASTKILL ? 1 : 0;
2130 
2131 	error = sysctl_int(oldp, oldlenp, newp, newlen, &flag);
2132 	if (error == 0 && newp) {
2133 		if (flag)
2134 			atomic_setbits_int(&findpr->ps_flags,
2135 			    PS_NOBROADCASTKILL);
2136 		else
2137 			atomic_clearbits_int(&findpr->ps_flags,
2138 			    PS_NOBROADCASTKILL);
2139 	}
2140 
2141 	return (error);
2142 }
2143 
2144 /* Arbitrary but reasonable limit for one iteration. */
2145 #define	VMMAP_MAXLEN	MAXPHYS
2146 
2147 int
2148 sysctl_proc_vmmap(int *name, u_int namelen, void *oldp, size_t *oldlenp,
2149     struct proc *cp)
2150 {
2151 	struct process *findpr;
2152 	pid_t pid;
2153 	int error;
2154 	size_t oldlen, len;
2155 	struct kinfo_vmentry *kve, *ukve;
2156 	u_long *ustart, start;
2157 
2158 	if (namelen > 1)
2159 		return (ENOTDIR);
2160 	if (namelen < 1)
2161 		return (EINVAL);
2162 
2163 	/* Provide max buffer length as hint. */
2164 	if (oldp == NULL) {
2165 		if (oldlenp == NULL)
2166 			return (EINVAL);
2167 		else {
2168 			*oldlenp = VMMAP_MAXLEN;
2169 			return (0);
2170 		}
2171 	}
2172 
2173 	pid = name[0];
2174 	if (pid == cp->p_p->ps_pid) {
2175 		/* Self process mapping. */
2176 		findpr = cp->p_p;
2177 	} else if (pid > 0) {
2178 		if ((findpr = prfind(pid)) == NULL)
2179 			return (ESRCH);
2180 
2181 		/* Either system process or exiting/zombie */
2182 		if (findpr->ps_flags & (PS_SYSTEM | PS_EXITING))
2183 			return (EINVAL);
2184 
2185 #if 1
2186 		/* XXX Allow only root for now */
2187 		if ((error = suser(cp)) != 0)
2188 			return (error);
2189 #else
2190 		/* Only owner or root can get vmmap */
2191 		if (findpr->ps_ucred->cr_uid != cp->p_ucred->cr_uid &&
2192 		    (error = suser(cp)) != 0)
2193 			return (error);
2194 #endif
2195 	} else {
2196 		/* Only root can get kernel_map */
2197 		if ((error = suser(cp)) != 0)
2198 			return (error);
2199 		findpr = NULL;
2200 	}
2201 
2202 	/* Check the given size. */
2203 	oldlen = *oldlenp;
2204 	if (oldlen == 0 || oldlen % sizeof(*kve) != 0)
2205 		return (EINVAL);
2206 
2207 	/* Deny huge allocation. */
2208 	if (oldlen > VMMAP_MAXLEN)
2209 		return (EINVAL);
2210 
2211 	/*
2212 	 * Iterate from the given address passed as the first element's
2213 	 * kve_start via oldp.
2214 	 */
2215 	ukve = (struct kinfo_vmentry *)oldp;
2216 	ustart = &ukve->kve_start;
2217 	error = copyin(ustart, &start, sizeof(start));
2218 	if (error != 0)
2219 		return (error);
2220 
2221 	/* Allocate wired memory to not block. */
2222 	kve = malloc(oldlen, M_TEMP, M_WAITOK);
2223 
2224 	/* Set the base address and read entries. */
2225 	kve[0].kve_start = start;
2226 	len = oldlen;
2227 	error = fill_vmmap(findpr, kve, &len);
2228 	if (error != 0 && error != ENOMEM)
2229 		goto done;
2230 	if (len == 0)
2231 		goto done;
2232 
2233 	KASSERT(len <= oldlen);
2234 	KASSERT((len % sizeof(struct kinfo_vmentry)) == 0);
2235 
2236 	error = copyout(kve, oldp, len);
2237 
2238 done:
2239 	*oldlenp = len;
2240 
2241 	free(kve, M_TEMP, oldlen);
2242 
2243 	return (error);
2244 }
2245 #endif
2246 
2247 /*
2248  * Initialize disknames/diskstats for export by sysctl. If update is set,
2249  * then we simply update the disk statistics information.
2250  */
2251 int
2252 sysctl_diskinit(int update, struct proc *p)
2253 {
2254 	struct diskstats *sdk;
2255 	struct disk *dk;
2256 	const char *duid;
2257 	int error, changed = 0;
2258 
2259 	KERNEL_ASSERT_LOCKED();
2260 
2261 	if ((error = rw_enter(&sysctl_disklock, RW_WRITE|RW_INTR)) != 0)
2262 		return error;
2263 
2264 	/* Run in a loop, disks may change while malloc sleeps. */
2265 	while (disk_change) {
2266 		int tlen;
2267 
2268 		disk_change = 0;
2269 
2270 		tlen = 0;
2271 		TAILQ_FOREACH(dk, &disklist, dk_link) {
2272 			if (dk->dk_name)
2273 				tlen += strlen(dk->dk_name);
2274 			tlen += 18;	/* label uid + separators */
2275 		}
2276 		tlen++;
2277 
2278 		/*
2279 		 * The sysctl_disklock ensures that no other process can
2280 		 * allocate disknames and diskstats while our malloc sleeps.
2281 		 */
2282 		free(disknames, M_SYSCTL, disknameslen);
2283 		free(diskstats, M_SYSCTL, diskstatslen);
2284 		diskstats = NULL;
2285 		disknames = NULL;
2286 		diskstats = mallocarray(disk_count, sizeof(struct diskstats),
2287 		    M_SYSCTL, M_WAITOK|M_ZERO);
2288 		diskstatslen = disk_count * sizeof(struct diskstats);
2289 		disknames = malloc(tlen, M_SYSCTL, M_WAITOK|M_ZERO);
2290 		disknameslen = tlen;
2291 		disknames[0] = '\0';
2292 		changed = 1;
2293 	}
2294 
2295 	if (changed) {
2296 		int l;
2297 
2298 		l = 0;
2299 		sdk = diskstats;
2300 		TAILQ_FOREACH(dk, &disklist, dk_link) {
2301 			duid = NULL;
2302 			if (dk->dk_label && !duid_iszero(dk->dk_label->d_uid))
2303 				duid = duid_format(dk->dk_label->d_uid);
2304 			snprintf(disknames + l, disknameslen - l, "%s:%s,",
2305 			    dk->dk_name ? dk->dk_name : "",
2306 			    duid ? duid : "");
2307 			l += strlen(disknames + l);
2308 			strlcpy(sdk->ds_name, dk->dk_name,
2309 			    sizeof(sdk->ds_name));
2310 			mtx_enter(&dk->dk_mtx);
2311 			sdk->ds_busy = dk->dk_busy;
2312 			sdk->ds_rxfer = dk->dk_rxfer;
2313 			sdk->ds_wxfer = dk->dk_wxfer;
2314 			sdk->ds_seek = dk->dk_seek;
2315 			sdk->ds_rbytes = dk->dk_rbytes;
2316 			sdk->ds_wbytes = dk->dk_wbytes;
2317 			sdk->ds_attachtime = dk->dk_attachtime;
2318 			sdk->ds_timestamp = dk->dk_timestamp;
2319 			sdk->ds_time = dk->dk_time;
2320 			mtx_leave(&dk->dk_mtx);
2321 			sdk++;
2322 		}
2323 
2324 		/* Eliminate trailing comma */
2325 		if (l != 0)
2326 			disknames[l - 1] = '\0';
2327 	} else if (update) {
2328 		/* Just update, number of drives hasn't changed */
2329 		sdk = diskstats;
2330 		TAILQ_FOREACH(dk, &disklist, dk_link) {
2331 			strlcpy(sdk->ds_name, dk->dk_name,
2332 			    sizeof(sdk->ds_name));
2333 			mtx_enter(&dk->dk_mtx);
2334 			sdk->ds_busy = dk->dk_busy;
2335 			sdk->ds_rxfer = dk->dk_rxfer;
2336 			sdk->ds_wxfer = dk->dk_wxfer;
2337 			sdk->ds_seek = dk->dk_seek;
2338 			sdk->ds_rbytes = dk->dk_rbytes;
2339 			sdk->ds_wbytes = dk->dk_wbytes;
2340 			sdk->ds_attachtime = dk->dk_attachtime;
2341 			sdk->ds_timestamp = dk->dk_timestamp;
2342 			sdk->ds_time = dk->dk_time;
2343 			mtx_leave(&dk->dk_mtx);
2344 			sdk++;
2345 		}
2346 	}
2347 	rw_exit_write(&sysctl_disklock);
2348 	return 0;
2349 }
2350 
2351 #if defined(SYSVMSG) || defined(SYSVSEM) || defined(SYSVSHM)
2352 int
2353 sysctl_sysvipc(int *name, u_int namelen, void *where, size_t *sizep)
2354 {
2355 #ifdef SYSVSEM
2356 	struct sem_sysctl_info *semsi;
2357 #endif
2358 #ifdef SYSVSHM
2359 	struct shm_sysctl_info *shmsi;
2360 #endif
2361 	size_t infosize, dssize, tsize, buflen, bufsiz;
2362 	int i, nds, error, ret;
2363 	void *buf;
2364 
2365 	if (namelen != 1)
2366 		return (EINVAL);
2367 
2368 	buflen = *sizep;
2369 
2370 	switch (*name) {
2371 	case KERN_SYSVIPC_MSG_INFO:
2372 #ifdef SYSVMSG
2373 		return (sysctl_sysvmsg(name, namelen, where, sizep));
2374 #else
2375 		return (EOPNOTSUPP);
2376 #endif
2377 	case KERN_SYSVIPC_SEM_INFO:
2378 #ifdef SYSVSEM
2379 		infosize = sizeof(semsi->seminfo);
2380 		nds = seminfo.semmni;
2381 		dssize = sizeof(semsi->semids[0]);
2382 		break;
2383 #else
2384 		return (EOPNOTSUPP);
2385 #endif
2386 	case KERN_SYSVIPC_SHM_INFO:
2387 #ifdef SYSVSHM
2388 		infosize = sizeof(shmsi->shminfo);
2389 		nds = shminfo.shmmni;
2390 		dssize = sizeof(shmsi->shmids[0]);
2391 		break;
2392 #else
2393 		return (EOPNOTSUPP);
2394 #endif
2395 	default:
2396 		return (EINVAL);
2397 	}
2398 	tsize = infosize + (nds * dssize);
2399 
2400 	/* Return just the total size required. */
2401 	if (where == NULL) {
2402 		*sizep = tsize;
2403 		return (0);
2404 	}
2405 
2406 	/* Not enough room for even the info struct. */
2407 	if (buflen < infosize) {
2408 		*sizep = 0;
2409 		return (ENOMEM);
2410 	}
2411 	bufsiz = min(tsize, buflen);
2412 	buf = malloc(bufsiz, M_TEMP, M_WAITOK|M_ZERO);
2413 
2414 	switch (*name) {
2415 #ifdef SYSVSEM
2416 	case KERN_SYSVIPC_SEM_INFO:
2417 		semsi = (struct sem_sysctl_info *)buf;
2418 		semsi->seminfo = seminfo;
2419 		break;
2420 #endif
2421 #ifdef SYSVSHM
2422 	case KERN_SYSVIPC_SHM_INFO:
2423 		shmsi = (struct shm_sysctl_info *)buf;
2424 		shmsi->shminfo = shminfo;
2425 		break;
2426 #endif
2427 	}
2428 	buflen -= infosize;
2429 
2430 	ret = 0;
2431 	if (buflen > 0) {
2432 		/* Fill in the IPC data structures.  */
2433 		for (i = 0; i < nds; i++) {
2434 			if (buflen < dssize) {
2435 				ret = ENOMEM;
2436 				break;
2437 			}
2438 			switch (*name) {
2439 #ifdef SYSVSEM
2440 			case KERN_SYSVIPC_SEM_INFO:
2441 				if (sema[i] != NULL)
2442 					memcpy(&semsi->semids[i], sema[i],
2443 					    dssize);
2444 				else
2445 					memset(&semsi->semids[i], 0, dssize);
2446 				break;
2447 #endif
2448 #ifdef SYSVSHM
2449 			case KERN_SYSVIPC_SHM_INFO:
2450 				if (shmsegs[i] != NULL)
2451 					memcpy(&shmsi->shmids[i], shmsegs[i],
2452 					    dssize);
2453 				else
2454 					memset(&shmsi->shmids[i], 0, dssize);
2455 				break;
2456 #endif
2457 			}
2458 			buflen -= dssize;
2459 		}
2460 	}
2461 	*sizep -= buflen;
2462 	error = copyout(buf, where, *sizep);
2463 	free(buf, M_TEMP, bufsiz);
2464 	/* If copyout succeeded, use return code set earlier. */
2465 	return (error ? error : ret);
2466 }
2467 #endif /* SYSVMSG || SYSVSEM || SYSVSHM */
2468 
2469 #ifndef	SMALL_KERNEL
2470 
2471 int
2472 sysctl_intrcnt(int *name, u_int namelen, void *oldp, size_t *oldlenp)
2473 {
2474 	return (evcount_sysctl(name, namelen, oldp, oldlenp, NULL, 0));
2475 }
2476 
2477 
2478 int
2479 sysctl_sensors(int *name, u_int namelen, void *oldp, size_t *oldlenp,
2480     void *newp, size_t newlen)
2481 {
2482 	struct ksensor *ks;
2483 	struct sensor *us;
2484 	struct ksensordev *ksd;
2485 	struct sensordev *usd;
2486 	int dev, numt, ret;
2487 	enum sensor_type type;
2488 
2489 	if (namelen != 1 && namelen != 3)
2490 		return (ENOTDIR);
2491 
2492 	dev = name[0];
2493 	if (namelen == 1) {
2494 		ret = sensordev_get(dev, &ksd);
2495 		if (ret)
2496 			return (ret);
2497 
2498 		/* Grab a copy, to clear the kernel pointers */
2499 		usd = malloc(sizeof(*usd), M_TEMP, M_WAITOK|M_ZERO);
2500 		usd->num = ksd->num;
2501 		strlcpy(usd->xname, ksd->xname, sizeof(usd->xname));
2502 		memcpy(usd->maxnumt, ksd->maxnumt, sizeof(usd->maxnumt));
2503 		usd->sensors_count = ksd->sensors_count;
2504 
2505 		ret = sysctl_rdstruct(oldp, oldlenp, newp, usd,
2506 		    sizeof(struct sensordev));
2507 
2508 		free(usd, M_TEMP, sizeof(*usd));
2509 		return (ret);
2510 	}
2511 
2512 	type = name[1];
2513 	numt = name[2];
2514 
2515 	ret = sensor_find(dev, type, numt, &ks);
2516 	if (ret)
2517 		return (ret);
2518 
2519 	/* Grab a copy, to clear the kernel pointers */
2520 	us = malloc(sizeof(*us), M_TEMP, M_WAITOK|M_ZERO);
2521 	memcpy(us->desc, ks->desc, sizeof(us->desc));
2522 	us->tv = ks->tv;
2523 	us->value = ks->value;
2524 	us->type = ks->type;
2525 	us->status = ks->status;
2526 	us->numt = ks->numt;
2527 	us->flags = ks->flags;
2528 
2529 	ret = sysctl_rdstruct(oldp, oldlenp, newp, us,
2530 	    sizeof(struct sensor));
2531 	free(us, M_TEMP, sizeof(*us));
2532 	return (ret);
2533 }
2534 #endif	/* SMALL_KERNEL */
2535 
2536 int
2537 sysctl_cptime2(int *name, u_int namelen, void *oldp, size_t *oldlenp,
2538     void *newp, size_t newlen)
2539 {
2540 	CPU_INFO_ITERATOR cii;
2541 	struct cpu_info *ci;
2542 	int found = 0;
2543 
2544 	if (namelen != 1)
2545 		return (ENOTDIR);
2546 
2547 	CPU_INFO_FOREACH(cii, ci) {
2548 		if (name[0] == CPU_INFO_UNIT(ci)) {
2549 			found = 1;
2550 			break;
2551 		}
2552 	}
2553 	if (!found)
2554 		return (ENOENT);
2555 
2556 	return (sysctl_rdstruct(oldp, oldlenp, newp,
2557 	    &ci->ci_schedstate.spc_cp_time,
2558 	    sizeof(ci->ci_schedstate.spc_cp_time)));
2559 }
2560 
2561 #if NAUDIO > 0
2562 int
2563 sysctl_audio(int *name, u_int namelen, void *oldp, size_t *oldlenp,
2564     void *newp, size_t newlen)
2565 {
2566 	if (namelen != 1)
2567 		return (ENOTDIR);
2568 
2569 	if (name[0] != KERN_AUDIO_RECORD)
2570 		return (ENOENT);
2571 
2572 	return (sysctl_int(oldp, oldlenp, newp, newlen, &audio_record_enable));
2573 }
2574 #endif
2575 
2576 #if NVIDEO > 0
2577 int
2578 sysctl_video(int *name, u_int namelen, void *oldp, size_t *oldlenp,
2579     void *newp, size_t newlen)
2580 {
2581 	if (namelen != 1)
2582 		return (ENOTDIR);
2583 
2584 	if (name[0] != KERN_VIDEO_RECORD)
2585 		return (ENOENT);
2586 
2587 	return (sysctl_int(oldp, oldlenp, newp, newlen, &video_record_enable));
2588 }
2589 #endif
2590 
2591 int
2592 sysctl_cpustats(int *name, u_int namelen, void *oldp, size_t *oldlenp,
2593     void *newp, size_t newlen)
2594 {
2595 	CPU_INFO_ITERATOR cii;
2596 	struct cpustats cs;
2597 	struct cpu_info *ci;
2598 	int found = 0;
2599 
2600 	if (namelen != 1)
2601 		return (ENOTDIR);
2602 
2603 	CPU_INFO_FOREACH(cii, ci) {
2604 		if (name[0] == CPU_INFO_UNIT(ci)) {
2605 			found = 1;
2606 			break;
2607 		}
2608 	}
2609 	if (!found)
2610 		return (ENOENT);
2611 
2612 	memset(&cs, 0, sizeof cs);
2613 	memcpy(&cs.cs_time, &ci->ci_schedstate.spc_cp_time, sizeof(cs.cs_time));
2614 	cs.cs_flags = 0;
2615 	if (cpu_is_online(ci))
2616 		cs.cs_flags |= CPUSTATS_ONLINE;
2617 
2618 	return (sysctl_rdstruct(oldp, oldlenp, newp, &cs, sizeof(cs)));
2619 }
2620 
2621 int
2622 sysctl_utc_offset(void *oldp, size_t *oldlenp, void *newp, size_t newlen)
2623 {
2624 	struct timespec adjusted, now;
2625 	int adjustment_seconds, error, new_offset_minutes, old_offset_minutes;
2626 
2627 	old_offset_minutes = utc_offset / 60;	/* seconds -> minutes */
2628 	new_offset_minutes = old_offset_minutes;
2629 	error = sysctl_securelevel_int(oldp, oldlenp, newp, newlen,
2630 	     &new_offset_minutes);
2631 	if (error)
2632 		return error;
2633 	if (new_offset_minutes < -24 * 60 || new_offset_minutes > 24 * 60)
2634 		return EINVAL;
2635 	if (new_offset_minutes == old_offset_minutes)
2636 		return 0;
2637 
2638 	utc_offset = new_offset_minutes * 60;	/* minutes -> seconds */
2639 	adjustment_seconds = (new_offset_minutes - old_offset_minutes) * 60;
2640 
2641 	nanotime(&now);
2642 	adjusted = now;
2643 	adjusted.tv_sec -= adjustment_seconds;
2644 	tc_setrealtimeclock(&adjusted);
2645 	resettodr();
2646 
2647 	return 0;
2648 }
2649