xref: /openbsd-src/sys/kern/kern_sysctl.c (revision f2da64fbbbf1b03f09f390ab01267c93dfd77c4c)
1 /*	$OpenBSD: kern_sysctl.c,v 1.310 2016/09/18 14:32:54 deraadt 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/file.h>
51 #include <sys/filedesc.h>
52 #include <sys/vnode.h>
53 #include <sys/unistd.h>
54 #include <sys/buf.h>
55 #include <sys/ioctl.h>
56 #include <sys/tty.h>
57 #include <sys/disklabel.h>
58 #include <sys/disk.h>
59 #include <sys/sysctl.h>
60 #include <sys/msgbuf.h>
61 #include <sys/vmmeter.h>
62 #include <sys/namei.h>
63 #include <sys/exec.h>
64 #include <sys/mbuf.h>
65 #include <sys/sensors.h>
66 #include <sys/pipe.h>
67 #include <sys/eventvar.h>
68 #include <sys/socketvar.h>
69 #include <sys/socket.h>
70 #include <sys/domain.h>
71 #include <sys/protosw.h>
72 #include <sys/pledge.h>
73 #include <sys/timetc.h>
74 #include <sys/evcount.h>
75 #include <sys/un.h>
76 #include <sys/unpcb.h>
77 #include <sys/sched.h>
78 #include <sys/mount.h>
79 #include <sys/syscallargs.h>
80 
81 #include <uvm/uvm_extern.h>
82 
83 #include <dev/cons.h>
84 #include <dev/rndvar.h>
85 
86 #include <net/route.h>
87 #include <netinet/in.h>
88 #include <netinet/ip.h>
89 #include <netinet/ip_var.h>
90 #include <netinet/in_pcb.h>
91 #include <netinet/ip6.h>
92 #include <netinet/tcp.h>
93 #include <netinet/tcp_timer.h>
94 #include <netinet/tcp_var.h>
95 #include <netinet/udp.h>
96 #include <netinet/udp_var.h>
97 #include <netinet6/ip6_var.h>
98 
99 #ifdef DDB
100 #include <ddb/db_var.h>
101 #endif
102 
103 #ifdef SYSVMSG
104 #include <sys/msg.h>
105 #endif
106 #ifdef SYSVSEM
107 #include <sys/sem.h>
108 #endif
109 #ifdef SYSVSHM
110 #include <sys/shm.h>
111 #endif
112 
113 extern struct forkstat forkstat;
114 extern struct nchstats nchstats;
115 extern int nselcoll, fscale;
116 extern struct disklist_head disklist;
117 extern fixpt_t ccpu;
118 extern  long numvnodes;
119 extern u_int net_livelocks;
120 
121 extern void nmbclust_update(void);
122 
123 int sysctl_diskinit(int, struct proc *);
124 int sysctl_proc_args(int *, u_int, void *, size_t *, struct proc *);
125 int sysctl_proc_cwd(int *, u_int, void *, size_t *, struct proc *);
126 int sysctl_proc_nobroadcastkill(int *, u_int, void *, size_t, void *, size_t *,
127 	struct proc *);
128 int sysctl_proc_vmmap(int *, u_int, void *, size_t *, struct proc *);
129 int sysctl_intrcnt(int *, u_int, void *, size_t *);
130 int sysctl_sensors(int *, u_int, void *, size_t *, void *, size_t);
131 int sysctl_cptime2(int *, u_int, void *, size_t *, void *, size_t);
132 
133 void fill_file(struct kinfo_file *, struct file *, struct filedesc *, int,
134     struct vnode *, struct process *, struct proc *, struct socket *, int);
135 void fill_kproc(struct process *, struct kinfo_proc *, struct proc *, int);
136 
137 int (*cpu_cpuspeed)(int *);
138 
139 /*
140  * Lock to avoid too many processes vslocking a large amount of memory
141  * at the same time.
142  */
143 struct rwlock sysctl_lock = RWLOCK_INITIALIZER("sysctllk");
144 struct rwlock sysctl_disklock = RWLOCK_INITIALIZER("sysctldlk");
145 
146 int
147 sys_sysctl(struct proc *p, void *v, register_t *retval)
148 {
149 	struct sys_sysctl_args /* {
150 		syscallarg(const int *) name;
151 		syscallarg(u_int) namelen;
152 		syscallarg(void *) old;
153 		syscallarg(size_t *) oldlenp;
154 		syscallarg(void *) new;
155 		syscallarg(size_t) newlen;
156 	} */ *uap = v;
157 	int error, dolock = 1;
158 	size_t savelen = 0, oldlen = 0;
159 	sysctlfn *fn;
160 	int name[CTL_MAXNAME];
161 
162 	if (SCARG(uap, new) != NULL &&
163 	    (error = suser(p, 0)))
164 		return (error);
165 	/*
166 	 * all top-level sysctl names are non-terminal
167 	 */
168 	if (SCARG(uap, namelen) > CTL_MAXNAME || SCARG(uap, namelen) < 2)
169 		return (EINVAL);
170 	error = copyin(SCARG(uap, name), name,
171 		       SCARG(uap, namelen) * sizeof(int));
172 	if (error)
173 		return (error);
174 
175 	error = pledge_sysctl(p, SCARG(uap, namelen),
176 	    name, SCARG(uap, new));
177 	if (error)
178 		return (error);
179 
180 	switch (name[0]) {
181 	case CTL_KERN:
182 		fn = kern_sysctl;
183 		break;
184 	case CTL_HW:
185 		fn = hw_sysctl;
186 		break;
187 	case CTL_VM:
188 		fn = uvm_sysctl;
189 		break;
190 	case CTL_NET:
191 		fn = net_sysctl;
192 		break;
193 	case CTL_FS:
194 		fn = fs_sysctl;
195 		break;
196 	case CTL_VFS:
197 		fn = vfs_sysctl;
198 		break;
199 	case CTL_MACHDEP:
200 		fn = cpu_sysctl;
201 		break;
202 #ifdef DEBUG
203 	case CTL_DEBUG:
204 		fn = debug_sysctl;
205 		break;
206 #endif
207 #ifdef DDB
208 	case CTL_DDB:
209 		fn = ddb_sysctl;
210 		break;
211 #endif
212 	default:
213 		return (EOPNOTSUPP);
214 	}
215 
216 	if (SCARG(uap, oldlenp) &&
217 	    (error = copyin(SCARG(uap, oldlenp), &oldlen, sizeof(oldlen))))
218 		return (error);
219 	if (SCARG(uap, old) != NULL) {
220 		if ((error = rw_enter(&sysctl_lock, RW_WRITE|RW_INTR)) != 0)
221 			return (error);
222 		if (dolock) {
223 			if (atop(oldlen) > uvmexp.wiredmax - uvmexp.wired) {
224 				rw_exit_write(&sysctl_lock);
225 				return (ENOMEM);
226 			}
227 			error = uvm_vslock(p, SCARG(uap, old), oldlen,
228 			    PROT_READ | PROT_WRITE);
229 			if (error) {
230 				rw_exit_write(&sysctl_lock);
231 				return (error);
232 			}
233 		}
234 		savelen = oldlen;
235 	}
236 	error = (*fn)(&name[1], SCARG(uap, namelen) - 1, SCARG(uap, old),
237 	    &oldlen, SCARG(uap, new), SCARG(uap, newlen), p);
238 	if (SCARG(uap, old) != NULL) {
239 		if (dolock)
240 			uvm_vsunlock(p, SCARG(uap, old), savelen);
241 		rw_exit_write(&sysctl_lock);
242 	}
243 	if (error)
244 		return (error);
245 	if (SCARG(uap, oldlenp))
246 		error = copyout(&oldlen, SCARG(uap, oldlenp), sizeof(oldlen));
247 	return (error);
248 }
249 
250 /*
251  * Attributes stored in the kernel.
252  */
253 char hostname[MAXHOSTNAMELEN];
254 int hostnamelen;
255 char domainname[MAXHOSTNAMELEN];
256 int domainnamelen;
257 long hostid;
258 char *disknames = NULL;
259 size_t disknameslen;
260 struct diskstats *diskstats = NULL;
261 size_t diskstatslen;
262 int securelevel;
263 
264 /*
265  * kernel related system variables.
266  */
267 int
268 kern_sysctl(int *name, u_int namelen, void *oldp, size_t *oldlenp, void *newp,
269     size_t newlen, struct proc *p)
270 {
271 	int error, level, inthostid, stackgap;
272 	dev_t dev;
273 	extern int somaxconn, sominconn;
274 	extern int nosuidcoredump;
275 	extern int maxlocksperuid;
276 	extern int pool_debug;
277 	extern int uvm_wxabort;
278 
279 	/* all sysctl names at this level are terminal except a ton of them */
280 	if (namelen != 1) {
281 		switch (name[0]) {
282 		case KERN_PROC:
283 		case KERN_PROF:
284 		case KERN_MALLOCSTATS:
285 		case KERN_TTY:
286 		case KERN_POOL:
287 		case KERN_PROC_ARGS:
288 		case KERN_PROC_CWD:
289 		case KERN_PROC_NOBROADCASTKILL:
290 		case KERN_PROC_VMMAP:
291 		case KERN_SYSVIPC_INFO:
292 		case KERN_SEMINFO:
293 		case KERN_SHMINFO:
294 		case KERN_INTRCNT:
295 		case KERN_WATCHDOG:
296 		case KERN_EVCOUNT:
297 		case KERN_TIMECOUNTER:
298 		case KERN_CPTIME2:
299 		case KERN_FILE:
300 			break;
301 		default:
302 			return (ENOTDIR);	/* overloaded */
303 		}
304 	}
305 
306 	switch (name[0]) {
307 	case KERN_OSTYPE:
308 		return (sysctl_rdstring(oldp, oldlenp, newp, ostype));
309 	case KERN_OSRELEASE:
310 		return (sysctl_rdstring(oldp, oldlenp, newp, osrelease));
311 	case KERN_OSREV:
312 		return (sysctl_rdint(oldp, oldlenp, newp, OpenBSD));
313 	case KERN_OSVERSION:
314 		return (sysctl_rdstring(oldp, oldlenp, newp, osversion));
315 	case KERN_VERSION:
316 		return (sysctl_rdstring(oldp, oldlenp, newp, version));
317 	case KERN_MAXVNODES:
318 		return(sysctl_int(oldp, oldlenp, newp, newlen, &maxvnodes));
319 	case KERN_MAXPROC:
320 		return (sysctl_int(oldp, oldlenp, newp, newlen, &maxprocess));
321 	case KERN_MAXFILES:
322 		return (sysctl_int(oldp, oldlenp, newp, newlen, &maxfiles));
323 	case KERN_NFILES:
324 		return (sysctl_rdint(oldp, oldlenp, newp, numfiles));
325 	case KERN_TTYCOUNT:
326 		return (sysctl_rdint(oldp, oldlenp, newp, tty_count));
327 	case KERN_NUMVNODES:
328 		return (sysctl_rdint(oldp, oldlenp, newp, numvnodes));
329 	case KERN_ARGMAX:
330 		return (sysctl_rdint(oldp, oldlenp, newp, ARG_MAX));
331 	case KERN_NSELCOLL:
332 		return (sysctl_rdint(oldp, oldlenp, newp, nselcoll));
333 	case KERN_SECURELVL:
334 		level = securelevel;
335 		if ((error = sysctl_int(oldp, oldlenp, newp, newlen, &level)) ||
336 		    newp == NULL)
337 			return (error);
338 		if ((securelevel > 0 || level < -1) &&
339 		    level < securelevel && p->p_p->ps_pid != 1)
340 			return (EPERM);
341 		securelevel = level;
342 		return (0);
343 	case KERN_HOSTNAME:
344 		error = sysctl_tstring(oldp, oldlenp, newp, newlen,
345 		    hostname, sizeof(hostname));
346 		if (newp && !error)
347 			hostnamelen = newlen;
348 		return (error);
349 	case KERN_DOMAINNAME:
350 		error = sysctl_tstring(oldp, oldlenp, newp, newlen,
351 		    domainname, sizeof(domainname));
352 		if (newp && !error)
353 			domainnamelen = newlen;
354 		return (error);
355 	case KERN_HOSTID:
356 		inthostid = hostid;  /* XXX assumes sizeof long <= sizeof int */
357 		error =  sysctl_int(oldp, oldlenp, newp, newlen, &inthostid);
358 		hostid = inthostid;
359 		return (error);
360 	case KERN_CLOCKRATE:
361 		return (sysctl_clockrate(oldp, oldlenp, newp));
362 	case KERN_BOOTTIME: {
363 		struct timeval bt;
364 		TIMESPEC_TO_TIMEVAL(&bt, &boottime);
365 		return (sysctl_rdstruct(oldp, oldlenp, newp, &bt, sizeof bt));
366 	  }
367 #ifndef SMALL_KERNEL
368 	case KERN_PROC:
369 		return (sysctl_doproc(name + 1, namelen - 1, oldp, oldlenp));
370 	case KERN_PROC_ARGS:
371 		return (sysctl_proc_args(name + 1, namelen - 1, oldp, oldlenp,
372 		     p));
373 	case KERN_PROC_CWD:
374 		return (sysctl_proc_cwd(name + 1, namelen - 1, oldp, oldlenp,
375 		     p));
376 	case KERN_PROC_NOBROADCASTKILL:
377 		return (sysctl_proc_nobroadcastkill(name + 1, namelen - 1,
378 		     newp, newlen, oldp, oldlenp, p));
379 	case KERN_PROC_VMMAP:
380 		return (sysctl_proc_vmmap(name + 1, namelen - 1, oldp, oldlenp,
381 		     p));
382 	case KERN_FILE:
383 		return (sysctl_file(name + 1, namelen - 1, oldp, oldlenp, p));
384 #endif
385 	case KERN_MBSTAT:
386 		return (sysctl_rdstruct(oldp, oldlenp, newp, &mbstat,
387 		    sizeof(mbstat)));
388 #if defined(GPROF) || defined(DDBPROF)
389 	case KERN_PROF:
390 		return (sysctl_doprof(name + 1, namelen - 1, oldp, oldlenp,
391 		    newp, newlen));
392 #endif
393 	case KERN_POSIX1:
394 		return (sysctl_rdint(oldp, oldlenp, newp, _POSIX_VERSION));
395 	case KERN_NGROUPS:
396 		return (sysctl_rdint(oldp, oldlenp, newp, NGROUPS_MAX));
397 	case KERN_JOB_CONTROL:
398 		return (sysctl_rdint(oldp, oldlenp, newp, 1));
399 	case KERN_SAVED_IDS:
400 		return (sysctl_rdint(oldp, oldlenp, newp, 1));
401 	case KERN_MAXPARTITIONS:
402 		return (sysctl_rdint(oldp, oldlenp, newp, MAXPARTITIONS));
403 	case KERN_RAWPARTITION:
404 		return (sysctl_rdint(oldp, oldlenp, newp, RAW_PART));
405 	case KERN_MAXTHREAD:
406 		return (sysctl_int(oldp, oldlenp, newp, newlen, &maxthread));
407 	case KERN_NTHREADS:
408 		return (sysctl_rdint(oldp, oldlenp, newp, nthreads));
409 	case KERN_SOMAXCONN:
410 		return (sysctl_int(oldp, oldlenp, newp, newlen, &somaxconn));
411 	case KERN_SOMINCONN:
412 		return (sysctl_int(oldp, oldlenp, newp, newlen, &sominconn));
413 	case KERN_ARND: {
414 		char buf[512];
415 
416 		if (*oldlenp > sizeof(buf))
417 			return (EINVAL);
418 		if (oldp) {
419 			arc4random_buf(buf, *oldlenp);
420 			if ((error = copyout(buf, oldp, *oldlenp)))
421 				return (error);
422 			explicit_bzero(buf, sizeof(buf));
423 		}
424 		return (0);
425 	}
426 	case KERN_NOSUIDCOREDUMP:
427 		return (sysctl_int(oldp, oldlenp, newp, newlen, &nosuidcoredump));
428 	case KERN_FSYNC:
429 		return (sysctl_rdint(oldp, oldlenp, newp, 1));
430 	case KERN_SYSVMSG:
431 #ifdef SYSVMSG
432 		return (sysctl_rdint(oldp, oldlenp, newp, 1));
433 #else
434 		return (sysctl_rdint(oldp, oldlenp, newp, 0));
435 #endif
436 	case KERN_SYSVSEM:
437 #ifdef SYSVSEM
438 		return (sysctl_rdint(oldp, oldlenp, newp, 1));
439 #else
440 		return (sysctl_rdint(oldp, oldlenp, newp, 0));
441 #endif
442 	case KERN_SYSVSHM:
443 #ifdef SYSVSHM
444 		return (sysctl_rdint(oldp, oldlenp, newp, 1));
445 #else
446 		return (sysctl_rdint(oldp, oldlenp, newp, 0));
447 #endif
448 	case KERN_MSGBUFSIZE:
449 	case KERN_CONSBUFSIZE: {
450 		struct msgbuf *mp;
451 		mp = (name[0] == KERN_MSGBUFSIZE) ? msgbufp : consbufp;
452 		/*
453 		 * deal with cases where the message buffer has
454 		 * become corrupted.
455 		 */
456 		if (!mp || mp->msg_magic != MSG_MAGIC)
457 			return (ENXIO);
458 		return (sysctl_rdint(oldp, oldlenp, newp, mp->msg_bufs));
459 	}
460 	case KERN_CONSBUF:
461 		if ((error = suser(p, 0)))
462 			return (error);
463 		/* FALLTHROUGH */
464 	case KERN_MSGBUF: {
465 		struct msgbuf *mp;
466 		mp = (name[0] == KERN_MSGBUF) ? msgbufp : consbufp;
467 		/* see note above */
468 		if (!mp || mp->msg_magic != MSG_MAGIC)
469 			return (ENXIO);
470 		return (sysctl_rdstruct(oldp, oldlenp, newp, mp,
471 		    mp->msg_bufs + offsetof(struct msgbuf, msg_bufc)));
472 	}
473 	case KERN_MALLOCSTATS:
474 		return (sysctl_malloc(name + 1, namelen - 1, oldp, oldlenp,
475 		    newp, newlen, p));
476 	case KERN_CPTIME:
477 	{
478 		CPU_INFO_ITERATOR cii;
479 		struct cpu_info *ci;
480 		long cp_time[CPUSTATES];
481 		int i;
482 
483 		memset(cp_time, 0, sizeof(cp_time));
484 
485 		CPU_INFO_FOREACH(cii, ci) {
486 			for (i = 0; i < CPUSTATES; i++)
487 				cp_time[i] += ci->ci_schedstate.spc_cp_time[i];
488 		}
489 
490 		for (i = 0; i < CPUSTATES; i++)
491 			cp_time[i] /= ncpus;
492 
493 		return (sysctl_rdstruct(oldp, oldlenp, newp, &cp_time,
494 		    sizeof(cp_time)));
495 	}
496 	case KERN_NCHSTATS:
497 		return (sysctl_rdstruct(oldp, oldlenp, newp, &nchstats,
498 		    sizeof(struct nchstats)));
499 	case KERN_FORKSTAT:
500 		return (sysctl_rdstruct(oldp, oldlenp, newp, &forkstat,
501 		    sizeof(struct forkstat)));
502 	case KERN_TTY:
503 		return (sysctl_tty(name + 1, namelen - 1, oldp, oldlenp,
504 		    newp, newlen));
505 	case KERN_FSCALE:
506 		return (sysctl_rdint(oldp, oldlenp, newp, fscale));
507 	case KERN_CCPU:
508 		return (sysctl_rdint(oldp, oldlenp, newp, ccpu));
509 	case KERN_NPROCS:
510 		return (sysctl_rdint(oldp, oldlenp, newp, nprocesses));
511 	case KERN_POOL:
512 		return (sysctl_dopool(name + 1, namelen - 1, oldp, oldlenp));
513 	case KERN_STACKGAPRANDOM:
514 		stackgap = stackgap_random;
515 		error = sysctl_int(oldp, oldlenp, newp, newlen, &stackgap);
516 		if (error)
517 			return (error);
518 		/*
519 		 * Safety harness.
520 		 */
521 		if ((stackgap < ALIGNBYTES && stackgap != 0) ||
522 		    !powerof2(stackgap) || stackgap >= MAXSSIZ)
523 			return (EINVAL);
524 		stackgap_random = stackgap;
525 		return (0);
526 #if defined(SYSVMSG) || defined(SYSVSEM) || defined(SYSVSHM)
527 	case KERN_SYSVIPC_INFO:
528 		return (sysctl_sysvipc(name + 1, namelen - 1, oldp, oldlenp));
529 #endif
530 	case KERN_SPLASSERT:
531 		return (sysctl_int(oldp, oldlenp, newp, newlen,
532 		    &splassert_ctl));
533 #ifdef SYSVSEM
534 	case KERN_SEMINFO:
535 		return (sysctl_sysvsem(name + 1, namelen - 1, oldp, oldlenp,
536 		    newp, newlen));
537 #endif
538 #ifdef SYSVSHM
539 	case KERN_SHMINFO:
540 		return (sysctl_sysvshm(name + 1, namelen - 1, oldp, oldlenp,
541 		    newp, newlen));
542 #endif
543 #ifndef SMALL_KERNEL
544 	case KERN_INTRCNT:
545 		return (sysctl_intrcnt(name + 1, namelen - 1, oldp, oldlenp));
546 	case KERN_WATCHDOG:
547 		return (sysctl_wdog(name + 1, namelen - 1, oldp, oldlenp,
548 		    newp, newlen));
549 #endif
550 	case KERN_MAXCLUSTERS:
551 		error = sysctl_int(oldp, oldlenp, newp, newlen, &nmbclust);
552 		if (!error)
553 			nmbclust_update();
554 		return (error);
555 #ifndef SMALL_KERNEL
556 	case KERN_EVCOUNT:
557 		return (evcount_sysctl(name + 1, namelen - 1, oldp, oldlenp,
558 		    newp, newlen));
559 #endif
560 	case KERN_TIMECOUNTER:
561 		return (sysctl_tc(name + 1, namelen - 1, oldp, oldlenp,
562 		    newp, newlen));
563 	case KERN_MAXLOCKSPERUID:
564 		return (sysctl_int(oldp, oldlenp, newp, newlen, &maxlocksperuid));
565 	case KERN_CPTIME2:
566 		return (sysctl_cptime2(name + 1, namelen -1, oldp, oldlenp,
567 		    newp, newlen));
568 	case KERN_CACHEPCT: {
569 		u_int64_t dmapages;
570 		int opct, pgs;
571 		opct = bufcachepercent;
572 		error = sysctl_int(oldp, oldlenp, newp, newlen,
573 		    &bufcachepercent);
574 		if (error)
575 			return(error);
576 		if (bufcachepercent > 90 || bufcachepercent < 5) {
577 			bufcachepercent = opct;
578 			return (EINVAL);
579 		}
580 		dmapages = uvm_pagecount(&dma_constraint);
581 		if (bufcachepercent != opct) {
582 			pgs = bufcachepercent * dmapages / 100;
583 			bufadjust(pgs); /* adjust bufpages */
584 			bufhighpages = bufpages; /* set high water mark */
585 		}
586 		return(0);
587 	}
588 	case KERN_WXABORT:
589 		return (sysctl_int(oldp, oldlenp, newp, newlen, &uvm_wxabort));
590 	case KERN_CONSDEV:
591 		if (cn_tab != NULL)
592 			dev = cn_tab->cn_dev;
593 		else
594 			dev = NODEV;
595 		return sysctl_rdstruct(oldp, oldlenp, newp, &dev, sizeof(dev));
596 	case KERN_NETLIVELOCKS:
597 		return (sysctl_rdint(oldp, oldlenp, newp, net_livelocks));
598 	case KERN_POOL_DEBUG: {
599 		int old_pool_debug = pool_debug;
600 
601 		error = sysctl_int(oldp, oldlenp, newp, newlen,
602 		    &pool_debug);
603 		if (error == 0 && pool_debug != old_pool_debug)
604 			pool_reclaim_all();
605 		return (error);
606 	}
607 #ifdef PTRACE
608 	case KERN_GLOBAL_PTRACE: {
609 		extern int global_ptrace;
610 
611 		return sysctl_int(oldp, oldlenp, newp, newlen, &global_ptrace);
612 	}
613 #endif
614 	default:
615 		return (EOPNOTSUPP);
616 	}
617 	/* NOTREACHED */
618 }
619 
620 /*
621  * hardware related system variables.
622  */
623 char *hw_vendor, *hw_prod, *hw_uuid, *hw_serial, *hw_ver;
624 int allowpowerdown = 1;
625 
626 int
627 hw_sysctl(int *name, u_int namelen, void *oldp, size_t *oldlenp, void *newp,
628     size_t newlen, struct proc *p)
629 {
630 	extern char machine[], cpu_model[];
631 	int err, cpuspeed;
632 
633 	/* all sysctl names at this level except sensors are terminal */
634 	if (name[0] != HW_SENSORS && namelen != 1)
635 		return (ENOTDIR);		/* overloaded */
636 
637 	switch (name[0]) {
638 	case HW_MACHINE:
639 		return (sysctl_rdstring(oldp, oldlenp, newp, machine));
640 	case HW_MODEL:
641 		return (sysctl_rdstring(oldp, oldlenp, newp, cpu_model));
642 	case HW_NCPU:
643 		return (sysctl_rdint(oldp, oldlenp, newp, ncpus));
644 	case HW_NCPUFOUND:
645 		return (sysctl_rdint(oldp, oldlenp, newp, ncpusfound));
646 	case HW_BYTEORDER:
647 		return (sysctl_rdint(oldp, oldlenp, newp, BYTE_ORDER));
648 	case HW_PHYSMEM:
649 		return (sysctl_rdint(oldp, oldlenp, newp, ptoa(physmem)));
650 	case HW_USERMEM:
651 		return (sysctl_rdint(oldp, oldlenp, newp,
652 		    ptoa(physmem - uvmexp.wired)));
653 	case HW_PAGESIZE:
654 		return (sysctl_rdint(oldp, oldlenp, newp, PAGE_SIZE));
655 	case HW_DISKNAMES:
656 		err = sysctl_diskinit(0, p);
657 		if (err)
658 			return err;
659 		if (disknames)
660 			return (sysctl_rdstring(oldp, oldlenp, newp,
661 			    disknames));
662 		else
663 			return (sysctl_rdstring(oldp, oldlenp, newp, ""));
664 	case HW_DISKSTATS:
665 		err = sysctl_diskinit(1, p);
666 		if (err)
667 			return err;
668 		return (sysctl_rdstruct(oldp, oldlenp, newp, diskstats,
669 		    disk_count * sizeof(struct diskstats)));
670 	case HW_DISKCOUNT:
671 		return (sysctl_rdint(oldp, oldlenp, newp, disk_count));
672 	case HW_CPUSPEED:
673 		if (!cpu_cpuspeed)
674 			return (EOPNOTSUPP);
675 		err = cpu_cpuspeed(&cpuspeed);
676 		if (err)
677 			return err;
678 		return (sysctl_rdint(oldp, oldlenp, newp, cpuspeed));
679 #ifndef	SMALL_KERNEL
680 	case HW_SENSORS:
681 		return (sysctl_sensors(name + 1, namelen - 1, oldp, oldlenp,
682 		    newp, newlen));
683 	case HW_SETPERF:
684 		return (sysctl_hwsetperf(oldp, oldlenp, newp, newlen));
685 	case HW_PERFPOLICY:
686 		return (sysctl_hwperfpolicy(oldp, oldlenp, newp, newlen));
687 #endif /* !SMALL_KERNEL */
688 	case HW_VENDOR:
689 		if (hw_vendor)
690 			return (sysctl_rdstring(oldp, oldlenp, newp,
691 			    hw_vendor));
692 		else
693 			return (EOPNOTSUPP);
694 	case HW_PRODUCT:
695 		if (hw_prod)
696 			return (sysctl_rdstring(oldp, oldlenp, newp, hw_prod));
697 		else
698 			return (EOPNOTSUPP);
699 	case HW_VERSION:
700 		if (hw_ver)
701 			return (sysctl_rdstring(oldp, oldlenp, newp, hw_ver));
702 		else
703 			return (EOPNOTSUPP);
704 	case HW_SERIALNO:
705 		if (hw_serial)
706 			return (sysctl_rdstring(oldp, oldlenp, newp,
707 			    hw_serial));
708 		else
709 			return (EOPNOTSUPP);
710 	case HW_UUID:
711 		if (hw_uuid)
712 			return (sysctl_rdstring(oldp, oldlenp, newp, hw_uuid));
713 		else
714 			return (EOPNOTSUPP);
715 	case HW_PHYSMEM64:
716 		return (sysctl_rdquad(oldp, oldlenp, newp,
717 		    ptoa((psize_t)physmem)));
718 	case HW_USERMEM64:
719 		return (sysctl_rdquad(oldp, oldlenp, newp,
720 		    ptoa((psize_t)physmem - uvmexp.wired)));
721 	case HW_ALLOWPOWERDOWN:
722 		if (securelevel > 0)
723 			return (sysctl_rdint(oldp, oldlenp, newp,
724 			    allowpowerdown));
725 		return (sysctl_int(oldp, oldlenp, newp, newlen,
726 		    &allowpowerdown));
727 	default:
728 		return (EOPNOTSUPP);
729 	}
730 	/* NOTREACHED */
731 }
732 
733 #ifdef DEBUG
734 /*
735  * Debugging related system variables.
736  */
737 extern struct ctldebug debug0, debug1;
738 struct ctldebug debug2, debug3, debug4;
739 struct ctldebug debug5, debug6, debug7, debug8, debug9;
740 struct ctldebug debug10, debug11, debug12, debug13, debug14;
741 struct ctldebug debug15, debug16, debug17, debug18, debug19;
742 static struct ctldebug *debugvars[CTL_DEBUG_MAXID] = {
743 	&debug0, &debug1, &debug2, &debug3, &debug4,
744 	&debug5, &debug6, &debug7, &debug8, &debug9,
745 	&debug10, &debug11, &debug12, &debug13, &debug14,
746 	&debug15, &debug16, &debug17, &debug18, &debug19,
747 };
748 int
749 debug_sysctl(int *name, u_int namelen, void *oldp, size_t *oldlenp, void *newp,
750     size_t newlen, struct proc *p)
751 {
752 	struct ctldebug *cdp;
753 
754 	/* all sysctl names at this level are name and field */
755 	if (namelen != 2)
756 		return (ENOTDIR);		/* overloaded */
757 	if (name[0] < 0 || name[0] >= nitems(debugvars))
758 		return (EOPNOTSUPP);
759 	cdp = debugvars[name[0]];
760 	if (cdp->debugname == 0)
761 		return (EOPNOTSUPP);
762 	switch (name[1]) {
763 	case CTL_DEBUG_NAME:
764 		return (sysctl_rdstring(oldp, oldlenp, newp, cdp->debugname));
765 	case CTL_DEBUG_VALUE:
766 		return (sysctl_int(oldp, oldlenp, newp, newlen, cdp->debugvar));
767 	default:
768 		return (EOPNOTSUPP);
769 	}
770 	/* NOTREACHED */
771 }
772 #endif /* DEBUG */
773 
774 /*
775  * Reads, or writes that lower the value
776  */
777 int
778 sysctl_int_lower(void *oldp, size_t *oldlenp, void *newp, size_t newlen, int *valp)
779 {
780 	unsigned int oval = *valp, val = *valp;
781 	int error;
782 
783 	if (newp == NULL)
784 		return (sysctl_rdint(oldp, oldlenp, newp, *valp));
785 
786 	if ((error = sysctl_int(oldp, oldlenp, newp, newlen, &val)))
787 		return (error);
788 	if (val > oval)
789 		return (EPERM);		/* do not allow raising */
790 	*(unsigned int *)valp = val;
791 	return (0);
792 }
793 
794 /*
795  * Validate parameters and get old / set new parameters
796  * for an integer-valued sysctl function.
797  */
798 int
799 sysctl_int(void *oldp, size_t *oldlenp, void *newp, size_t newlen, int *valp)
800 {
801 	int error = 0;
802 
803 	if (oldp && *oldlenp < sizeof(int))
804 		return (ENOMEM);
805 	if (newp && newlen != sizeof(int))
806 		return (EINVAL);
807 	*oldlenp = sizeof(int);
808 	if (oldp)
809 		error = copyout(valp, oldp, sizeof(int));
810 	if (error == 0 && newp)
811 		error = copyin(newp, valp, sizeof(int));
812 	return (error);
813 }
814 
815 /*
816  * As above, but read-only.
817  */
818 int
819 sysctl_rdint(void *oldp, size_t *oldlenp, void *newp, int val)
820 {
821 	int error = 0;
822 
823 	if (oldp && *oldlenp < sizeof(int))
824 		return (ENOMEM);
825 	if (newp)
826 		return (EPERM);
827 	*oldlenp = sizeof(int);
828 	if (oldp)
829 		error = copyout((caddr_t)&val, oldp, sizeof(int));
830 	return (error);
831 }
832 
833 /*
834  * Array of integer values.
835  */
836 int
837 sysctl_int_arr(int **valpp, int *name, u_int namelen, void *oldp,
838     size_t *oldlenp, void *newp, size_t newlen)
839 {
840 	if (namelen > 1)
841 		return (ENOTDIR);
842 	if (name[0] < 0 || valpp[name[0]] == NULL)
843 		return (EOPNOTSUPP);
844 	return (sysctl_int(oldp, oldlenp, newp, newlen, valpp[name[0]]));
845 }
846 
847 /*
848  * Validate parameters and get old / set new parameters
849  * for an integer-valued sysctl function.
850  */
851 int
852 sysctl_quad(void *oldp, size_t *oldlenp, void *newp, size_t newlen,
853     int64_t *valp)
854 {
855 	int error = 0;
856 
857 	if (oldp && *oldlenp < sizeof(int64_t))
858 		return (ENOMEM);
859 	if (newp && newlen != sizeof(int64_t))
860 		return (EINVAL);
861 	*oldlenp = sizeof(int64_t);
862 	if (oldp)
863 		error = copyout(valp, oldp, sizeof(int64_t));
864 	if (error == 0 && newp)
865 		error = copyin(newp, valp, sizeof(int64_t));
866 	return (error);
867 }
868 
869 /*
870  * As above, but read-only.
871  */
872 int
873 sysctl_rdquad(void *oldp, size_t *oldlenp, void *newp, int64_t val)
874 {
875 	int error = 0;
876 
877 	if (oldp && *oldlenp < sizeof(int64_t))
878 		return (ENOMEM);
879 	if (newp)
880 		return (EPERM);
881 	*oldlenp = sizeof(int64_t);
882 	if (oldp)
883 		error = copyout((caddr_t)&val, oldp, sizeof(int64_t));
884 	return (error);
885 }
886 
887 /*
888  * Validate parameters and get old / set new parameters
889  * for a string-valued sysctl function.
890  */
891 int
892 sysctl_string(void *oldp, size_t *oldlenp, void *newp, size_t newlen, char *str,
893     int maxlen)
894 {
895 	return sysctl__string(oldp, oldlenp, newp, newlen, str, maxlen, 0);
896 }
897 
898 int
899 sysctl_tstring(void *oldp, size_t *oldlenp, void *newp, size_t newlen,
900     char *str, int maxlen)
901 {
902 	return sysctl__string(oldp, oldlenp, newp, newlen, str, maxlen, 1);
903 }
904 
905 int
906 sysctl__string(void *oldp, size_t *oldlenp, void *newp, size_t newlen,
907     char *str, int maxlen, int trunc)
908 {
909 	int len, error = 0;
910 
911 	len = strlen(str) + 1;
912 	if (oldp && *oldlenp < len) {
913 		if (trunc == 0 || *oldlenp == 0)
914 			return (ENOMEM);
915 	}
916 	if (newp && newlen >= maxlen)
917 		return (EINVAL);
918 	if (oldp) {
919 		if (trunc && *oldlenp < len) {
920 			len = *oldlenp;
921 			error = copyout(str, oldp, len - 1);
922 			if (error == 0)
923 				error = copyout("", (char *)oldp + len - 1, 1);
924 		} else {
925 			error = copyout(str, oldp, len);
926 		}
927 	}
928 	*oldlenp = len;
929 	if (error == 0 && newp) {
930 		error = copyin(newp, str, newlen);
931 		str[newlen] = 0;
932 	}
933 	return (error);
934 }
935 
936 /*
937  * As above, but read-only.
938  */
939 int
940 sysctl_rdstring(void *oldp, size_t *oldlenp, void *newp, const char *str)
941 {
942 	int len, error = 0;
943 
944 	len = strlen(str) + 1;
945 	if (oldp && *oldlenp < len)
946 		return (ENOMEM);
947 	if (newp)
948 		return (EPERM);
949 	*oldlenp = len;
950 	if (oldp)
951 		error = copyout(str, oldp, len);
952 	return (error);
953 }
954 
955 /*
956  * Validate parameters and get old / set new parameters
957  * for a structure oriented sysctl function.
958  */
959 int
960 sysctl_struct(void *oldp, size_t *oldlenp, void *newp, size_t newlen, void *sp,
961     int len)
962 {
963 	int error = 0;
964 
965 	if (oldp && *oldlenp < len)
966 		return (ENOMEM);
967 	if (newp && newlen > len)
968 		return (EINVAL);
969 	if (oldp) {
970 		*oldlenp = len;
971 		error = copyout(sp, oldp, len);
972 	}
973 	if (error == 0 && newp)
974 		error = copyin(newp, sp, len);
975 	return (error);
976 }
977 
978 /*
979  * Validate parameters and get old parameters
980  * for a structure oriented sysctl function.
981  */
982 int
983 sysctl_rdstruct(void *oldp, size_t *oldlenp, void *newp, const void *sp,
984     int len)
985 {
986 	int error = 0;
987 
988 	if (oldp && *oldlenp < len)
989 		return (ENOMEM);
990 	if (newp)
991 		return (EPERM);
992 	*oldlenp = len;
993 	if (oldp)
994 		error = copyout(sp, oldp, len);
995 	return (error);
996 }
997 
998 #ifndef SMALL_KERNEL
999 void
1000 fill_file(struct kinfo_file *kf, struct file *fp, struct filedesc *fdp,
1001 	  int fd, struct vnode *vp, struct process *pr, struct proc *p,
1002 	  struct socket *so, int show_pointers)
1003 {
1004 	struct vattr va;
1005 
1006 	memset(kf, 0, sizeof(*kf));
1007 
1008 	kf->fd_fd = fd;		/* might not really be an fd */
1009 
1010 	if (fp != NULL) {
1011 		if (show_pointers)
1012 			kf->f_fileaddr = PTRTOINT64(fp);
1013 		kf->f_flag = fp->f_flag;
1014 		kf->f_iflags = fp->f_iflags;
1015 		kf->f_type = fp->f_type;
1016 		kf->f_count = fp->f_count;
1017 		if (show_pointers)
1018 			kf->f_ucred = PTRTOINT64(fp->f_cred);
1019 		kf->f_uid = fp->f_cred->cr_uid;
1020 		kf->f_gid = fp->f_cred->cr_gid;
1021 		if (show_pointers)
1022 			kf->f_ops = PTRTOINT64(fp->f_ops);
1023 		if (show_pointers)
1024 			kf->f_data = PTRTOINT64(fp->f_data);
1025 		kf->f_usecount = 0;
1026 
1027 		if (suser(p, 0) == 0 || p->p_ucred->cr_uid == fp->f_cred->cr_uid) {
1028 			kf->f_offset = fp->f_offset;
1029 			kf->f_rxfer = fp->f_rxfer;
1030 			kf->f_rwfer = fp->f_wxfer;
1031 			kf->f_seek = fp->f_seek;
1032 			kf->f_rbytes = fp->f_rbytes;
1033 			kf->f_wbytes = fp->f_wbytes;
1034 		} else
1035 			kf->f_offset = -1;
1036 	} else if (vp != NULL) {
1037 		/* fake it */
1038 		kf->f_type = DTYPE_VNODE;
1039 		kf->f_flag = FREAD;
1040 		if (fd == KERN_FILE_TRACE)
1041 			kf->f_flag |= FWRITE;
1042 	} else if (so != NULL) {
1043 		/* fake it */
1044 		kf->f_type = DTYPE_SOCKET;
1045 	}
1046 
1047 	/* information about the object associated with this file */
1048 	switch (kf->f_type) {
1049 	case DTYPE_VNODE:
1050 		if (fp != NULL)
1051 			vp = (struct vnode *)fp->f_data;
1052 
1053 		if (show_pointers)
1054 			kf->v_un = PTRTOINT64(vp->v_un.vu_socket);
1055 		kf->v_type = vp->v_type;
1056 		kf->v_tag = vp->v_tag;
1057 		kf->v_flag = vp->v_flag;
1058 		if (show_pointers)
1059 			kf->v_data = PTRTOINT64(vp->v_data);
1060 		if (show_pointers)
1061 			kf->v_mount = PTRTOINT64(vp->v_mount);
1062 		if (vp->v_mount)
1063 			strlcpy(kf->f_mntonname,
1064 			    vp->v_mount->mnt_stat.f_mntonname,
1065 			    sizeof(kf->f_mntonname));
1066 
1067 		if (VOP_GETATTR(vp, &va, p->p_ucred, p) == 0) {
1068 			kf->va_fileid = va.va_fileid;
1069 			kf->va_mode = MAKEIMODE(va.va_type, va.va_mode);
1070 			kf->va_size = va.va_size;
1071 			kf->va_rdev = va.va_rdev;
1072 			kf->va_fsid = va.va_fsid & 0xffffffff;
1073 		}
1074 		break;
1075 
1076 	case DTYPE_SOCKET: {
1077 		if (so == NULL)
1078 			so = (struct socket *)fp->f_data;
1079 
1080 		kf->so_type = so->so_type;
1081 		kf->so_state = so->so_state;
1082 		if (show_pointers)
1083 			kf->so_pcb = PTRTOINT64(so->so_pcb);
1084 		else
1085 			kf->so_pcb = -1;
1086 		kf->so_protocol = so->so_proto->pr_protocol;
1087 		kf->so_family = so->so_proto->pr_domain->dom_family;
1088 		kf->so_rcv_cc = so->so_rcv.sb_cc;
1089 		kf->so_snd_cc = so->so_snd.sb_cc;
1090 		if (isspliced(so)) {
1091 			if (show_pointers)
1092 				kf->so_splice =
1093 				    PTRTOINT64(so->so_sp->ssp_socket);
1094 			kf->so_splicelen = so->so_sp->ssp_len;
1095 		} else if (issplicedback(so))
1096 			kf->so_splicelen = -1;
1097 		if (!so->so_pcb)
1098 			break;
1099 		switch (kf->so_family) {
1100 		case AF_INET: {
1101 			struct inpcb *inpcb = so->so_pcb;
1102 
1103 			if (show_pointers)
1104 				kf->inp_ppcb = PTRTOINT64(inpcb->inp_ppcb);
1105 			kf->inp_lport = inpcb->inp_lport;
1106 			kf->inp_laddru[0] = inpcb->inp_laddr.s_addr;
1107 			kf->inp_fport = inpcb->inp_fport;
1108 			kf->inp_faddru[0] = inpcb->inp_faddr.s_addr;
1109 			kf->inp_rtableid = inpcb->inp_rtableid;
1110 			if (so->so_type == SOCK_RAW)
1111 				kf->inp_proto = inpcb->inp_ip.ip_p;
1112 			if (so->so_proto->pr_protocol == IPPROTO_TCP) {
1113 				struct tcpcb *tcpcb = (void *)inpcb->inp_ppcb;
1114 				kf->t_rcv_wnd = tcpcb->rcv_wnd;
1115 				kf->t_snd_wnd = tcpcb->snd_wnd;
1116 				kf->t_snd_cwnd = tcpcb->snd_cwnd;
1117 				kf->t_state = tcpcb->t_state;
1118 			}
1119 			break;
1120 		    }
1121 		case AF_INET6: {
1122 			struct inpcb *inpcb = so->so_pcb;
1123 
1124 			kf->inp_ppcb = PTRTOINT64(inpcb->inp_ppcb);
1125 			kf->inp_lport = inpcb->inp_lport;
1126 			kf->inp_laddru[0] = inpcb->inp_laddr6.s6_addr32[0];
1127 			kf->inp_laddru[1] = inpcb->inp_laddr6.s6_addr32[1];
1128 			kf->inp_laddru[2] = inpcb->inp_laddr6.s6_addr32[2];
1129 			kf->inp_laddru[3] = inpcb->inp_laddr6.s6_addr32[3];
1130 			kf->inp_fport = inpcb->inp_fport;
1131 			kf->inp_faddru[0] = inpcb->inp_faddr6.s6_addr32[0];
1132 			kf->inp_faddru[1] = inpcb->inp_faddr6.s6_addr32[1];
1133 			kf->inp_faddru[2] = inpcb->inp_faddr6.s6_addr32[2];
1134 			kf->inp_faddru[3] = inpcb->inp_faddr6.s6_addr32[3];
1135 			kf->inp_rtableid = inpcb->inp_rtableid;
1136 			if (so->so_type == SOCK_RAW)
1137 				kf->inp_proto = inpcb->inp_ipv6.ip6_nxt;
1138 			if (so->so_proto->pr_protocol == IPPROTO_TCP) {
1139 				struct tcpcb *tcpcb = (void *)inpcb->inp_ppcb;
1140 				kf->t_rcv_wnd = tcpcb->rcv_wnd;
1141 				kf->t_snd_wnd = tcpcb->snd_wnd;
1142 				kf->t_state = tcpcb->t_state;
1143 			}
1144 			break;
1145 		    }
1146 		case AF_UNIX: {
1147 			struct unpcb *unpcb = so->so_pcb;
1148 
1149 			kf->f_msgcount = unpcb->unp_msgcount;
1150 			if (show_pointers) {
1151 				kf->unp_conn	= PTRTOINT64(unpcb->unp_conn);
1152 				kf->unp_refs	= PTRTOINT64(
1153 				    SLIST_FIRST(&unpcb->unp_refs));
1154 				kf->unp_nextref	= PTRTOINT64(
1155 				    SLIST_NEXT(unpcb, unp_nextref));
1156 				kf->v_un	= PTRTOINT64(unpcb->unp_vnode);
1157 				kf->unp_addr	= PTRTOINT64(unpcb->unp_addr);
1158 			}
1159 			if (unpcb->unp_addr != NULL) {
1160 				struct sockaddr_un *un = mtod(unpcb->unp_addr,
1161 				    struct sockaddr_un *);
1162 				memcpy(kf->unp_path, un->sun_path, un->sun_len
1163 				    - offsetof(struct sockaddr_un,sun_path));
1164 			}
1165 			break;
1166 		    }
1167 		}
1168 		break;
1169 	    }
1170 
1171 	case DTYPE_PIPE: {
1172 		struct pipe *pipe = (struct pipe *)fp->f_data;
1173 
1174 		if (show_pointers)
1175 			kf->pipe_peer = PTRTOINT64(pipe->pipe_peer);
1176 		kf->pipe_state = pipe->pipe_state;
1177 		break;
1178 	    }
1179 
1180 	case DTYPE_KQUEUE: {
1181 		struct kqueue *kqi = (struct kqueue *)fp->f_data;
1182 
1183 		kf->kq_count = kqi->kq_count;
1184 		kf->kq_state = kqi->kq_state;
1185 		break;
1186 	    }
1187 	}
1188 
1189 	/* per-process information for KERN_FILE_BY[PU]ID */
1190 	if (pr != NULL) {
1191 		kf->p_pid = pr->ps_pid;
1192 		kf->p_uid = pr->ps_ucred->cr_uid;
1193 		kf->p_gid = pr->ps_ucred->cr_gid;
1194 		kf->p_tid = -1;
1195 		strlcpy(kf->p_comm, pr->ps_mainproc->p_comm,
1196 		    sizeof(kf->p_comm));
1197 	}
1198 	if (fdp != NULL)
1199 		kf->fd_ofileflags = fdp->fd_ofileflags[fd];
1200 }
1201 
1202 /*
1203  * Get file structures.
1204  */
1205 int
1206 sysctl_file(int *name, u_int namelen, char *where, size_t *sizep,
1207     struct proc *p)
1208 {
1209 	struct kinfo_file *kf;
1210 	struct filedesc *fdp;
1211 	struct file *fp, *nfp;
1212 	struct process *pr;
1213 	size_t buflen, elem_size, elem_count, outsize;
1214 	char *dp = where;
1215 	int arg, i, error = 0, needed = 0, matched;
1216 	u_int op;
1217 	int show_pointers;
1218 
1219 	if (namelen > 4)
1220 		return (ENOTDIR);
1221 	if (namelen < 4 || name[2] > sizeof(*kf))
1222 		return (EINVAL);
1223 
1224 	buflen = where != NULL ? *sizep : 0;
1225 	op = name[0];
1226 	arg = name[1];
1227 	elem_size = name[2];
1228 	elem_count = name[3];
1229 	outsize = MIN(sizeof(*kf), elem_size);
1230 
1231 	if (elem_size < 1)
1232 		return (EINVAL);
1233 
1234 	show_pointers = suser(curproc, 0) == 0;
1235 
1236 	kf = malloc(sizeof(*kf), M_TEMP, M_WAITOK);
1237 
1238 #define FILLIT2(fp, fdp, i, vp, pr, so) do {				\
1239 	if (buflen >= elem_size && elem_count > 0) {			\
1240 		fill_file(kf, fp, fdp, i, vp, pr, p, so, show_pointers);\
1241 		error = copyout(kf, dp, outsize);			\
1242 		if (error)						\
1243 			break;						\
1244 		dp += elem_size;					\
1245 		buflen -= elem_size;					\
1246 		elem_count--;						\
1247 	}								\
1248 	needed += elem_size;						\
1249 } while (0)
1250 #define FILLIT(fp, fdp, i, vp, pr) \
1251 	FILLIT2(fp, fdp, i, vp, pr, NULL)
1252 #define FILLSO(so) \
1253 	FILLIT2(NULL, NULL, 0, NULL, NULL, so)
1254 
1255 	switch (op) {
1256 	case KERN_FILE_BYFILE:
1257 		/* use the inp-tables to pick up closed connections, too */
1258 		if (arg == DTYPE_SOCKET) {
1259 			extern struct inpcbtable rawcbtable;
1260 #ifdef INET6
1261 			extern struct inpcbtable rawin6pcbtable;
1262 #endif
1263 			struct inpcb *inp;
1264 			int s;
1265 
1266 			s = splnet();
1267 			TAILQ_FOREACH(inp, &tcbtable.inpt_queue, inp_queue)
1268 				FILLSO(inp->inp_socket);
1269 			TAILQ_FOREACH(inp, &udbtable.inpt_queue, inp_queue)
1270 				FILLSO(inp->inp_socket);
1271 			TAILQ_FOREACH(inp, &rawcbtable.inpt_queue, inp_queue)
1272 				FILLSO(inp->inp_socket);
1273 #ifdef INET6
1274 			TAILQ_FOREACH(inp, &rawin6pcbtable.inpt_queue,
1275 			    inp_queue)
1276 				FILLSO(inp->inp_socket);
1277 #endif
1278 			splx(s);
1279 		}
1280 		fp = LIST_FIRST(&filehead);
1281 		/* don't FREF when f_count == 0 to avoid race in fdrop() */
1282 		while (fp != NULL && fp->f_count == 0)
1283 			fp = LIST_NEXT(fp, f_list);
1284 		if (fp == NULL)
1285 			break;
1286 		FREF(fp);
1287 		do {
1288 			if (fp->f_count > 1 && /* 0, +1 for our FREF() */
1289 			    (arg == 0 || fp->f_type == arg)) {
1290 				int af, skip = 0;
1291 				if (arg == DTYPE_SOCKET && fp->f_type == arg) {
1292 					af = ((struct socket *)fp->f_data)->
1293 					    so_proto->pr_domain->dom_family;
1294 					if (af == AF_INET || af == AF_INET6)
1295 						skip = 1;
1296 				}
1297 				if (!skip)
1298 					FILLIT(fp, NULL, 0, NULL, NULL);
1299 			}
1300 			nfp = LIST_NEXT(fp, f_list);
1301 			while (nfp != NULL && nfp->f_count == 0)
1302 				nfp = LIST_NEXT(nfp, f_list);
1303 			if (nfp != NULL)
1304 				FREF(nfp);
1305 			FRELE(fp, p);
1306 			fp = nfp;
1307 		} while (fp != NULL);
1308 		break;
1309 	case KERN_FILE_BYPID:
1310 		/* A arg of -1 indicates all processes */
1311 		if (arg < -1) {
1312 			error = EINVAL;
1313 			break;
1314 		}
1315 		matched = 0;
1316 		LIST_FOREACH(pr, &allprocess, ps_list) {
1317 			/*
1318 			 * skip system, exiting, embryonic and undead
1319 			 * processes
1320 			 */
1321 			if (pr->ps_flags & (PS_SYSTEM | PS_EMBRYO | PS_EXITING))
1322 				continue;
1323 			if (arg > 0 && pr->ps_pid != (pid_t)arg) {
1324 				/* not the pid we are looking for */
1325 				continue;
1326 			}
1327 			matched = 1;
1328 			fdp = pr->ps_fd;
1329 			if (pr->ps_textvp)
1330 				FILLIT(NULL, NULL, KERN_FILE_TEXT, pr->ps_textvp, pr);
1331 			if (fdp->fd_cdir)
1332 				FILLIT(NULL, NULL, KERN_FILE_CDIR, fdp->fd_cdir, pr);
1333 			if (fdp->fd_rdir)
1334 				FILLIT(NULL, NULL, KERN_FILE_RDIR, fdp->fd_rdir, pr);
1335 			if (pr->ps_tracevp)
1336 				FILLIT(NULL, NULL, KERN_FILE_TRACE, pr->ps_tracevp, pr);
1337 			for (i = 0; i < fdp->fd_nfiles; i++) {
1338 				if ((fp = fdp->fd_ofiles[i]) == NULL)
1339 					continue;
1340 				if (!FILE_IS_USABLE(fp))
1341 					continue;
1342 				FILLIT(fp, fdp, i, NULL, pr);
1343 			}
1344 		}
1345 		if (!matched)
1346 			error = ESRCH;
1347 		break;
1348 	case KERN_FILE_BYUID:
1349 		LIST_FOREACH(pr, &allprocess, ps_list) {
1350 			/*
1351 			 * skip system, exiting, embryonic and undead
1352 			 * processes
1353 			 */
1354 			if (pr->ps_flags & (PS_SYSTEM | PS_EMBRYO | PS_EXITING))
1355 				continue;
1356 			if (arg >= 0 && pr->ps_ucred->cr_uid != (uid_t)arg) {
1357 				/* not the uid we are looking for */
1358 				continue;
1359 			}
1360 			fdp = pr->ps_fd;
1361 			if (fdp->fd_cdir)
1362 				FILLIT(NULL, NULL, KERN_FILE_CDIR, fdp->fd_cdir, pr);
1363 			if (fdp->fd_rdir)
1364 				FILLIT(NULL, NULL, KERN_FILE_RDIR, fdp->fd_rdir, pr);
1365 			if (pr->ps_tracevp)
1366 				FILLIT(NULL, NULL, KERN_FILE_TRACE, pr->ps_tracevp, pr);
1367 			for (i = 0; i < fdp->fd_nfiles; i++) {
1368 				if ((fp = fdp->fd_ofiles[i]) == NULL)
1369 					continue;
1370 				if (!FILE_IS_USABLE(fp))
1371 					continue;
1372 				FILLIT(fp, fdp, i, NULL, pr);
1373 			}
1374 		}
1375 		break;
1376 	default:
1377 		error = EINVAL;
1378 		break;
1379 	}
1380 	free(kf, M_TEMP, sizeof(*kf));
1381 
1382 	if (!error) {
1383 		if (where == NULL)
1384 			needed += KERN_FILESLOP * elem_size;
1385 		else if (*sizep < needed)
1386 			error = ENOMEM;
1387 		*sizep = needed;
1388 	}
1389 
1390 	return (error);
1391 }
1392 
1393 /*
1394  * try over estimating by 5 procs
1395  */
1396 #define KERN_PROCSLOP	5
1397 
1398 int
1399 sysctl_doproc(int *name, u_int namelen, char *where, size_t *sizep)
1400 {
1401 	struct kinfo_proc *kproc = NULL;
1402 	struct proc *p;
1403 	struct process *pr;
1404 	char *dp;
1405 	int arg, buflen, doingzomb, elem_size, elem_count;
1406 	int error, needed, op;
1407 	int dothreads = 0;
1408 	int show_pointers;
1409 
1410 	dp = where;
1411 	buflen = where != NULL ? *sizep : 0;
1412 	needed = error = 0;
1413 
1414 	if (namelen != 4 || name[2] < 0 || name[3] < 0 ||
1415 	    name[2] > sizeof(*kproc))
1416 		return (EINVAL);
1417 	op = name[0];
1418 	arg = name[1];
1419 	elem_size = name[2];
1420 	elem_count = name[3];
1421 
1422 	dothreads = op & KERN_PROC_SHOW_THREADS;
1423 	op &= ~KERN_PROC_SHOW_THREADS;
1424 
1425 	show_pointers = suser(curproc, 0) == 0;
1426 
1427 	if (where != NULL)
1428 		kproc = malloc(sizeof(*kproc), M_TEMP, M_WAITOK);
1429 
1430 	pr = LIST_FIRST(&allprocess);
1431 	doingzomb = 0;
1432 again:
1433 	for (; pr != NULL; pr = LIST_NEXT(pr, ps_list)) {
1434 		/* XXX skip processes in the middle of being zapped */
1435 		if (pr->ps_pgrp == NULL)
1436 			continue;
1437 
1438 		/*
1439 		 * Skip embryonic processes.
1440 		 */
1441 		if (pr->ps_flags & PS_EMBRYO)
1442 			continue;
1443 
1444 		/*
1445 		 * TODO - make more efficient (see notes below).
1446 		 */
1447 		switch (op) {
1448 
1449 		case KERN_PROC_PID:
1450 			/* could do this with just a lookup */
1451 			if (pr->ps_pid != (pid_t)arg)
1452 				continue;
1453 			break;
1454 
1455 		case KERN_PROC_PGRP:
1456 			/* could do this by traversing pgrp */
1457 			if (pr->ps_pgrp->pg_id != (pid_t)arg)
1458 				continue;
1459 			break;
1460 
1461 		case KERN_PROC_SESSION:
1462 			if (pr->ps_session->s_leader == NULL ||
1463 			    pr->ps_session->s_leader->ps_pid != (pid_t)arg)
1464 				continue;
1465 			break;
1466 
1467 		case KERN_PROC_TTY:
1468 			if ((pr->ps_flags & PS_CONTROLT) == 0 ||
1469 			    pr->ps_session->s_ttyp == NULL ||
1470 			    pr->ps_session->s_ttyp->t_dev != (dev_t)arg)
1471 				continue;
1472 			break;
1473 
1474 		case KERN_PROC_UID:
1475 			if (pr->ps_ucred->cr_uid != (uid_t)arg)
1476 				continue;
1477 			break;
1478 
1479 		case KERN_PROC_RUID:
1480 			if (pr->ps_ucred->cr_ruid != (uid_t)arg)
1481 				continue;
1482 			break;
1483 
1484 		case KERN_PROC_ALL:
1485 			if (pr->ps_flags & PS_SYSTEM)
1486 				continue;
1487 			break;
1488 
1489 		case KERN_PROC_KTHREAD:
1490 			/* no filtering */
1491 			break;
1492 
1493 		default:
1494 			error = EINVAL;
1495 			goto err;
1496 		}
1497 
1498 		if (buflen >= elem_size && elem_count > 0) {
1499 			fill_kproc(pr, kproc, NULL, show_pointers);
1500 			error = copyout(kproc, dp, elem_size);
1501 			if (error)
1502 				goto err;
1503 			dp += elem_size;
1504 			buflen -= elem_size;
1505 			elem_count--;
1506 		}
1507 		needed += elem_size;
1508 
1509 		/* Skip per-thread entries if not required by op */
1510 		if (!dothreads)
1511 			continue;
1512 
1513 		TAILQ_FOREACH(p, &pr->ps_threads, p_thr_link) {
1514 			if (buflen >= elem_size && elem_count > 0) {
1515 				fill_kproc(pr, kproc, p, show_pointers);
1516 				error = copyout(kproc, dp, elem_size);
1517 				if (error)
1518 					goto err;
1519 				dp += elem_size;
1520 				buflen -= elem_size;
1521 				elem_count--;
1522 			}
1523 			needed += elem_size;
1524 		}
1525 	}
1526 	if (doingzomb == 0) {
1527 		pr = LIST_FIRST(&zombprocess);
1528 		doingzomb++;
1529 		goto again;
1530 	}
1531 	if (where != NULL) {
1532 		*sizep = dp - where;
1533 		if (needed > *sizep) {
1534 			error = ENOMEM;
1535 			goto err;
1536 		}
1537 	} else {
1538 		needed += KERN_PROCSLOP * elem_size;
1539 		*sizep = needed;
1540 	}
1541 err:
1542 	if (kproc)
1543 		free(kproc, M_TEMP, sizeof(*kproc));
1544 	return (error);
1545 }
1546 
1547 /*
1548  * Fill in a kproc structure for the specified process.
1549  */
1550 void
1551 fill_kproc(struct process *pr, struct kinfo_proc *ki, struct proc *p,
1552     int show_pointers)
1553 {
1554 	struct session *s = pr->ps_session;
1555 	struct tty *tp;
1556 	struct timespec ut, st;
1557 	int isthread;
1558 
1559 	isthread = p != NULL;
1560 	if (!isthread)
1561 		p = pr->ps_mainproc;		/* XXX */
1562 
1563 	FILL_KPROC(ki, strlcpy, p, pr, pr->ps_ucred, pr->ps_pgrp,
1564 	    p, pr, s, pr->ps_vmspace, pr->ps_limit, pr->ps_sigacts, isthread,
1565 	    show_pointers);
1566 
1567 	/* stuff that's too painful to generalize into the macros */
1568 	ki->p_pid = pr->ps_pid;
1569 	if (pr->ps_pptr)
1570 		ki->p_ppid = pr->ps_pptr->ps_pid;
1571 	if (s->s_leader)
1572 		ki->p_sid = s->s_leader->ps_pid;
1573 
1574 	if ((pr->ps_flags & PS_CONTROLT) && (tp = s->s_ttyp)) {
1575 		ki->p_tdev = tp->t_dev;
1576 		ki->p_tpgid = tp->t_pgrp ? tp->t_pgrp->pg_id : -1;
1577 		if (show_pointers)
1578 			ki->p_tsess = PTRTOINT64(tp->t_session);
1579 	} else {
1580 		ki->p_tdev = NODEV;
1581 		ki->p_tpgid = -1;
1582 	}
1583 
1584 	/* fixups that can only be done in the kernel */
1585 	if ((pr->ps_flags & PS_ZOMBIE) == 0) {
1586 		if ((pr->ps_flags & PS_EMBRYO) == 0)
1587 			ki->p_vm_rssize = vm_resident_count(pr->ps_vmspace);
1588 		calctsru(isthread ? &p->p_tu : &pr->ps_tu, &ut, &st, NULL);
1589 		ki->p_uutime_sec = ut.tv_sec;
1590 		ki->p_uutime_usec = ut.tv_nsec/1000;
1591 		ki->p_ustime_sec = st.tv_sec;
1592 		ki->p_ustime_usec = st.tv_nsec/1000;
1593 
1594 #ifdef MULTIPROCESSOR
1595 		if (isthread && p->p_cpu != NULL)
1596 			ki->p_cpuid = CPU_INFO_UNIT(p->p_cpu);
1597 #endif
1598 	}
1599 
1600 	/* get %cpu and schedule state: just one thread or sum of all? */
1601 	if (isthread) {
1602 		ki->p_pctcpu = p->p_pctcpu;
1603 		ki->p_stat   = p->p_stat;
1604 	} else {
1605 		ki->p_pctcpu = 0;
1606 		ki->p_stat = (pr->ps_flags & PS_ZOMBIE) ? SDEAD : SIDL;
1607 		TAILQ_FOREACH(p, &pr->ps_threads, p_thr_link) {
1608 			ki->p_pctcpu += p->p_pctcpu;
1609 			/* find best state: ONPROC > RUN > STOP > SLEEP > .. */
1610 			if (p->p_stat == SONPROC || ki->p_stat == SONPROC)
1611 				ki->p_stat = SONPROC;
1612 			else if (p->p_stat == SRUN || ki->p_stat == SRUN)
1613 				ki->p_stat = SRUN;
1614 			else if (p->p_stat == SSTOP || ki->p_stat == SSTOP)
1615 				ki->p_stat = SSTOP;
1616 			else if (p->p_stat == SSLEEP)
1617 				ki->p_stat = SSLEEP;
1618 		}
1619 	}
1620 }
1621 
1622 int
1623 sysctl_proc_args(int *name, u_int namelen, void *oldp, size_t *oldlenp,
1624     struct proc *cp)
1625 {
1626 	struct process *vpr;
1627 	pid_t pid;
1628 	struct ps_strings pss;
1629 	struct iovec iov;
1630 	struct uio uio;
1631 	int error, cnt, op;
1632 	size_t limit;
1633 	char **rargv, **vargv;		/* reader vs. victim */
1634 	char *rarg, *varg, *buf;
1635 	struct vmspace *vm;
1636 	vaddr_t ps_strings;
1637 
1638 	if (namelen > 2)
1639 		return (ENOTDIR);
1640 	if (namelen < 2)
1641 		return (EINVAL);
1642 
1643 	pid = name[0];
1644 	op = name[1];
1645 
1646 	switch (op) {
1647 	case KERN_PROC_ARGV:
1648 	case KERN_PROC_NARGV:
1649 	case KERN_PROC_ENV:
1650 	case KERN_PROC_NENV:
1651 		break;
1652 	default:
1653 		return (EOPNOTSUPP);
1654 	}
1655 
1656 	if ((vpr = prfind(pid)) == NULL)
1657 		return (ESRCH);
1658 
1659 	if (oldp == NULL) {
1660 		if (op == KERN_PROC_NARGV || op == KERN_PROC_NENV)
1661 			*oldlenp = sizeof(int);
1662 		else
1663 			*oldlenp = ARG_MAX;	/* XXX XXX XXX */
1664 		return (0);
1665 	}
1666 
1667 	/* Either system process or exiting/zombie */
1668 	if (vpr->ps_flags & (PS_SYSTEM | PS_EXITING))
1669 		return (EINVAL);
1670 
1671 	/* Execing - danger. */
1672 	if ((vpr->ps_flags & PS_INEXEC))
1673 		return (EBUSY);
1674 
1675 	/* Only owner or root can get env */
1676 	if ((op == KERN_PROC_NENV || op == KERN_PROC_ENV) &&
1677 	    (vpr->ps_ucred->cr_uid != cp->p_ucred->cr_uid &&
1678 	    (error = suser(cp, 0)) != 0))
1679 		return (error);
1680 
1681 	ps_strings = vpr->ps_strings;
1682 	vm = vpr->ps_vmspace;
1683 	vm->vm_refcnt++;
1684 	vpr = NULL;
1685 
1686 	buf = malloc(PAGE_SIZE, M_TEMP, M_WAITOK);
1687 
1688 	iov.iov_base = &pss;
1689 	iov.iov_len = sizeof(pss);
1690 	uio.uio_iov = &iov;
1691 	uio.uio_iovcnt = 1;
1692 	uio.uio_offset = (off_t)ps_strings;
1693 	uio.uio_resid = sizeof(pss);
1694 	uio.uio_segflg = UIO_SYSSPACE;
1695 	uio.uio_rw = UIO_READ;
1696 	uio.uio_procp = cp;
1697 
1698 	if ((error = uvm_io(&vm->vm_map, &uio, 0)) != 0)
1699 		goto out;
1700 
1701 	if (op == KERN_PROC_NARGV) {
1702 		error = sysctl_rdint(oldp, oldlenp, NULL, pss.ps_nargvstr);
1703 		goto out;
1704 	}
1705 	if (op == KERN_PROC_NENV) {
1706 		error = sysctl_rdint(oldp, oldlenp, NULL, pss.ps_nenvstr);
1707 		goto out;
1708 	}
1709 
1710 	if (op == KERN_PROC_ARGV) {
1711 		cnt = pss.ps_nargvstr;
1712 		vargv = pss.ps_argvstr;
1713 	} else {
1714 		cnt = pss.ps_nenvstr;
1715 		vargv = pss.ps_envstr;
1716 	}
1717 
1718 	/* -1 to have space for a terminating NUL */
1719 	limit = *oldlenp - 1;
1720 	*oldlenp = 0;
1721 
1722 	rargv = oldp;
1723 
1724 	/*
1725 	 * *oldlenp - number of bytes copied out into readers buffer.
1726 	 * limit - maximal number of bytes allowed into readers buffer.
1727 	 * rarg - pointer into readers buffer where next arg will be stored.
1728 	 * rargv - pointer into readers buffer where the next rarg pointer
1729 	 *  will be stored.
1730 	 * vargv - pointer into victim address space where the next argument
1731 	 *  will be read.
1732 	 */
1733 
1734 	/* space for cnt pointers and a NULL */
1735 	rarg = (char *)(rargv + cnt + 1);
1736 	*oldlenp += (cnt + 1) * sizeof(char **);
1737 
1738 	while (cnt > 0 && *oldlenp < limit) {
1739 		size_t len, vstrlen;
1740 
1741 		/* Write to readers argv */
1742 		if ((error = copyout(&rarg, rargv, sizeof(rarg))) != 0)
1743 			goto out;
1744 
1745 		/* read the victim argv */
1746 		iov.iov_base = &varg;
1747 		iov.iov_len = sizeof(varg);
1748 		uio.uio_iov = &iov;
1749 		uio.uio_iovcnt = 1;
1750 		uio.uio_offset = (off_t)(vaddr_t)vargv;
1751 		uio.uio_resid = sizeof(varg);
1752 		uio.uio_segflg = UIO_SYSSPACE;
1753 		uio.uio_rw = UIO_READ;
1754 		uio.uio_procp = cp;
1755 		if ((error = uvm_io(&vm->vm_map, &uio, 0)) != 0)
1756 			goto out;
1757 
1758 		if (varg == NULL)
1759 			break;
1760 
1761 		/*
1762 		 * read the victim arg. We must jump through hoops to avoid
1763 		 * crossing a page boundary too much and returning an error.
1764 		 */
1765 more:
1766 		len = PAGE_SIZE - (((vaddr_t)varg) & PAGE_MASK);
1767 		/* leave space for the terminating NUL */
1768 		iov.iov_base = buf;
1769 		iov.iov_len = len;
1770 		uio.uio_iov = &iov;
1771 		uio.uio_iovcnt = 1;
1772 		uio.uio_offset = (off_t)(vaddr_t)varg;
1773 		uio.uio_resid = len;
1774 		uio.uio_segflg = UIO_SYSSPACE;
1775 		uio.uio_rw = UIO_READ;
1776 		uio.uio_procp = cp;
1777 		if ((error = uvm_io(&vm->vm_map, &uio, 0)) != 0)
1778 			goto out;
1779 
1780 		for (vstrlen = 0; vstrlen < len; vstrlen++) {
1781 			if (buf[vstrlen] == '\0')
1782 				break;
1783 		}
1784 
1785 		/* Don't overflow readers buffer. */
1786 		if (*oldlenp + vstrlen + 1 >= limit) {
1787 			error = ENOMEM;
1788 			goto out;
1789 		}
1790 
1791 		if ((error = copyout(buf, rarg, vstrlen)) != 0)
1792 			goto out;
1793 
1794 		*oldlenp += vstrlen;
1795 		rarg += vstrlen;
1796 
1797 		/* The string didn't end in this page? */
1798 		if (vstrlen == len) {
1799 			varg += vstrlen;
1800 			goto more;
1801 		}
1802 
1803 		/* End of string. Terminate it with a NUL */
1804 		buf[0] = '\0';
1805 		if ((error = copyout(buf, rarg, 1)) != 0)
1806 			goto out;
1807 		*oldlenp += 1;
1808 		rarg += 1;
1809 
1810 		vargv++;
1811 		rargv++;
1812 		cnt--;
1813 	}
1814 
1815 	if (*oldlenp >= limit) {
1816 		error = ENOMEM;
1817 		goto out;
1818 	}
1819 
1820 	/* Write the terminating null */
1821 	rarg = NULL;
1822 	error = copyout(&rarg, rargv, sizeof(rarg));
1823 
1824 out:
1825 	uvmspace_free(vm);
1826 	free(buf, M_TEMP, PAGE_SIZE);
1827 	return (error);
1828 }
1829 
1830 int
1831 sysctl_proc_cwd(int *name, u_int namelen, void *oldp, size_t *oldlenp,
1832     struct proc *cp)
1833 {
1834 	struct process *findpr;
1835 	struct vnode *vp;
1836 	pid_t pid;
1837 	int error;
1838 	size_t lenused, len;
1839 	char *path, *bp, *bend;
1840 
1841 	if (namelen > 1)
1842 		return (ENOTDIR);
1843 	if (namelen < 1)
1844 		return (EINVAL);
1845 
1846 	pid = name[0];
1847 	if ((findpr = prfind(pid)) == NULL)
1848 		return (ESRCH);
1849 
1850 	if (oldp == NULL) {
1851 		*oldlenp = MAXPATHLEN * 4;
1852 		return (0);
1853 	}
1854 
1855 	/* Either system process or exiting/zombie */
1856 	if (findpr->ps_flags & (PS_SYSTEM | PS_EXITING))
1857 		return (EINVAL);
1858 
1859 	/* Only owner or root can get cwd */
1860 	if (findpr->ps_ucred->cr_uid != cp->p_ucred->cr_uid &&
1861 	    (error = suser(cp, 0)) != 0)
1862 		return (error);
1863 
1864 	len = *oldlenp;
1865 	if (len > MAXPATHLEN * 4)
1866 		len = MAXPATHLEN * 4;
1867 	else if (len < 2)
1868 		return (ERANGE);
1869 	*oldlenp = 0;
1870 
1871 	/* snag a reference to the vnode before we can sleep */
1872 	vp = findpr->ps_fd->fd_cdir;
1873 	vref(vp);
1874 
1875 	path = malloc(len, M_TEMP, M_WAITOK);
1876 
1877 	bp = &path[len];
1878 	bend = bp;
1879 	*(--bp) = '\0';
1880 
1881 	/* Same as sys__getcwd */
1882 	error = vfs_getcwd_common(vp, NULL,
1883 	    &bp, path, len / 2, GETCWD_CHECK_ACCESS, cp);
1884 	if (error == 0) {
1885 		*oldlenp = lenused = bend - bp;
1886 		error = copyout(bp, oldp, lenused);
1887 	}
1888 
1889 	vrele(vp);
1890 	free(path, M_TEMP, len);
1891 
1892 	return (error);
1893 }
1894 
1895 int
1896 sysctl_proc_nobroadcastkill(int *name, u_int namelen, void *newp, size_t newlen,
1897     void *oldp, size_t *oldlenp, struct proc *cp)
1898 {
1899 	struct process *findpr;
1900 	pid_t pid;
1901 	int error, flag;
1902 
1903 	if (namelen > 1)
1904 		return (ENOTDIR);
1905 	if (namelen < 1)
1906 		return (EINVAL);
1907 
1908 	pid = name[0];
1909 	if ((findpr = prfind(pid)) == NULL)
1910 		return (ESRCH);
1911 
1912 	/* Either system process or exiting/zombie */
1913 	if (findpr->ps_flags & (PS_SYSTEM | PS_EXITING))
1914 		return (EINVAL);
1915 
1916 	/* Only root can change PS_NOBROADCASTKILL */
1917 	if (newp != 0 && (error = suser(cp, 0)) != 0)
1918 		return (error);
1919 
1920 	/* get the PS_NOBROADCASTKILL flag */
1921 	flag = findpr->ps_flags & PS_NOBROADCASTKILL ? 1 : 0;
1922 
1923 	error = sysctl_int(oldp, oldlenp, newp, newlen, &flag);
1924 	if (error == 0 && newp) {
1925 		if (flag)
1926 			atomic_setbits_int(&findpr->ps_flags,
1927 			    PS_NOBROADCASTKILL);
1928 		else
1929 			atomic_clearbits_int(&findpr->ps_flags,
1930 			    PS_NOBROADCASTKILL);
1931 	}
1932 
1933 	return (error);
1934 }
1935 
1936 /* Arbitrary but reasonable limit for one iteration. */
1937 #define	VMMAP_MAXLEN	MAXPHYS
1938 
1939 int
1940 sysctl_proc_vmmap(int *name, u_int namelen, void *oldp, size_t *oldlenp,
1941     struct proc *cp)
1942 {
1943 	struct process *findpr;
1944 	pid_t pid;
1945 	int error;
1946 	size_t oldlen, len;
1947 	struct kinfo_vmentry *kve, *ukve;
1948 	u_long *ustart, start;
1949 
1950 	if (namelen > 1)
1951 		return (ENOTDIR);
1952 	if (namelen < 1)
1953 		return (EINVAL);
1954 
1955 	/* Provide max buffer length as hint. */
1956 	if (oldp == NULL) {
1957 		if (oldlenp == NULL)
1958 			return (EINVAL);
1959 		else {
1960 			*oldlenp = VMMAP_MAXLEN;
1961 			return (0);
1962 		}
1963 	}
1964 
1965 	pid = name[0];
1966 	if (pid == cp->p_pid) {
1967 		/* Self process mapping. */
1968 		findpr = cp->p_p;
1969 	} else if (pid > 0) {
1970 		if ((findpr = prfind(pid)) == NULL)
1971 			return (ESRCH);
1972 
1973 		/* Either system process or exiting/zombie */
1974 		if (findpr->ps_flags & (PS_SYSTEM | PS_EXITING))
1975 			return (EINVAL);
1976 
1977 #if 1
1978 		/* XXX Allow only root for now */
1979 		if ((error = suser(cp, 0)) != 0)
1980 			return (error);
1981 #else
1982 		/* Only owner or root can get vmmap */
1983 		if (findpr->ps_ucred->cr_uid != cp->p_ucred->cr_uid &&
1984 		    (error = suser(cp, 0)) != 0)
1985 			return (error);
1986 #endif
1987 	} else {
1988 		/* Only root can get kernel_map */
1989 		if ((error = suser(cp, 0)) != 0)
1990 			return (error);
1991 		findpr = NULL;
1992 	}
1993 
1994 	/* Check the given size. */
1995 	oldlen = *oldlenp;
1996 	if (oldlen == 0 || oldlen % sizeof(*kve) != 0)
1997 		return (EINVAL);
1998 
1999 	/* Deny huge allocation. */
2000 	if (oldlen > VMMAP_MAXLEN)
2001 		return (EINVAL);
2002 
2003 	/*
2004 	 * Iterate from the given address passed as the first element's
2005 	 * kve_start via oldp.
2006 	 */
2007 	ukve = (struct kinfo_vmentry *)oldp;
2008 	ustart = &ukve->kve_start;
2009 	error = copyin(ustart, &start, sizeof(start));
2010 	if (error != 0)
2011 		return (error);
2012 
2013 	/* Allocate wired memory to not block. */
2014 	kve = malloc(oldlen, M_TEMP, M_WAITOK);
2015 
2016 	/* Set the base address and read entries. */
2017 	kve[0].kve_start = start;
2018 	len = oldlen;
2019 	error = fill_vmmap(findpr, kve, &len);
2020 	if (error != 0 && error != ENOMEM)
2021 		goto done;
2022 	if (len == 0)
2023 		goto done;
2024 
2025 	KASSERT(len <= oldlen);
2026 	KASSERT((len % sizeof(struct kinfo_vmentry)) == 0);
2027 
2028 	error = copyout(kve, oldp, len);
2029 
2030 done:
2031 	*oldlenp = len;
2032 
2033 	free(kve, M_TEMP, oldlen);
2034 
2035 	return (error);
2036 }
2037 #endif
2038 
2039 /*
2040  * Initialize disknames/diskstats for export by sysctl. If update is set,
2041  * then we simply update the disk statistics information.
2042  */
2043 int
2044 sysctl_diskinit(int update, struct proc *p)
2045 {
2046 	struct diskstats *sdk;
2047 	struct disk *dk;
2048 	const char *duid;
2049 	int i, tlen, l;
2050 
2051 	if ((i = rw_enter(&sysctl_disklock, RW_WRITE|RW_INTR)) != 0)
2052 		return i;
2053 
2054 	if (disk_change) {
2055 		for (dk = TAILQ_FIRST(&disklist), tlen = 0; dk;
2056 		    dk = TAILQ_NEXT(dk, dk_link)) {
2057 			if (dk->dk_name)
2058 				tlen += strlen(dk->dk_name);
2059 			tlen += 18;	/* label uid + separators */
2060 		}
2061 		tlen++;
2062 
2063 		if (disknames)
2064 			free(disknames, M_SYSCTL, disknameslen);
2065 		if (diskstats)
2066 			free(diskstats, M_SYSCTL, diskstatslen);
2067 		diskstats = NULL;
2068 		disknames = NULL;
2069 		diskstats = mallocarray(disk_count, sizeof(struct diskstats),
2070 		    M_SYSCTL, M_WAITOK);
2071 		diskstatslen = disk_count * sizeof(struct diskstats);
2072 		disknames = malloc(tlen, M_SYSCTL, M_WAITOK);
2073 		disknameslen = tlen;
2074 		disknames[0] = '\0';
2075 
2076 		for (dk = TAILQ_FIRST(&disklist), i = 0, l = 0; dk;
2077 		    dk = TAILQ_NEXT(dk, dk_link), i++) {
2078 			duid = NULL;
2079 			if (dk->dk_label && !duid_iszero(dk->dk_label->d_uid))
2080 				duid = duid_format(dk->dk_label->d_uid);
2081 			snprintf(disknames + l, tlen - l, "%s:%s,",
2082 			    dk->dk_name ? dk->dk_name : "",
2083 			    duid ? duid : "");
2084 			l += strlen(disknames + l);
2085 			sdk = diskstats + i;
2086 			strlcpy(sdk->ds_name, dk->dk_name,
2087 			    sizeof(sdk->ds_name));
2088 			mtx_enter(&dk->dk_mtx);
2089 			sdk->ds_busy = dk->dk_busy;
2090 			sdk->ds_rxfer = dk->dk_rxfer;
2091 			sdk->ds_wxfer = dk->dk_wxfer;
2092 			sdk->ds_seek = dk->dk_seek;
2093 			sdk->ds_rbytes = dk->dk_rbytes;
2094 			sdk->ds_wbytes = dk->dk_wbytes;
2095 			sdk->ds_attachtime = dk->dk_attachtime;
2096 			sdk->ds_timestamp = dk->dk_timestamp;
2097 			sdk->ds_time = dk->dk_time;
2098 			mtx_leave(&dk->dk_mtx);
2099 		}
2100 
2101 		/* Eliminate trailing comma */
2102 		if (l != 0)
2103 			disknames[l - 1] = '\0';
2104 		disk_change = 0;
2105 	} else if (update) {
2106 		/* Just update, number of drives hasn't changed */
2107 		for (dk = TAILQ_FIRST(&disklist), i = 0; dk;
2108 		    dk = TAILQ_NEXT(dk, dk_link), i++) {
2109 			sdk = diskstats + i;
2110 			strlcpy(sdk->ds_name, dk->dk_name,
2111 			    sizeof(sdk->ds_name));
2112 			mtx_enter(&dk->dk_mtx);
2113 			sdk->ds_busy = dk->dk_busy;
2114 			sdk->ds_rxfer = dk->dk_rxfer;
2115 			sdk->ds_wxfer = dk->dk_wxfer;
2116 			sdk->ds_seek = dk->dk_seek;
2117 			sdk->ds_rbytes = dk->dk_rbytes;
2118 			sdk->ds_wbytes = dk->dk_wbytes;
2119 			sdk->ds_attachtime = dk->dk_attachtime;
2120 			sdk->ds_timestamp = dk->dk_timestamp;
2121 			sdk->ds_time = dk->dk_time;
2122 			mtx_leave(&dk->dk_mtx);
2123 		}
2124 	}
2125 	rw_exit_write(&sysctl_disklock);
2126 	return 0;
2127 }
2128 
2129 #if defined(SYSVMSG) || defined(SYSVSEM) || defined(SYSVSHM)
2130 int
2131 sysctl_sysvipc(int *name, u_int namelen, void *where, size_t *sizep)
2132 {
2133 #ifdef SYSVSEM
2134 	struct sem_sysctl_info *semsi;
2135 #endif
2136 #ifdef SYSVSHM
2137 	struct shm_sysctl_info *shmsi;
2138 #endif
2139 	size_t infosize, dssize, tsize, buflen, bufsiz;
2140 	int i, nds, error, ret;
2141 	void *buf;
2142 
2143 	if (namelen != 1)
2144 		return (EINVAL);
2145 
2146 	buflen = *sizep;
2147 
2148 	switch (*name) {
2149 	case KERN_SYSVIPC_MSG_INFO:
2150 #ifdef SYSVMSG
2151 		return (sysctl_sysvmsg(name, namelen, where, sizep));
2152 #else
2153 		return (EOPNOTSUPP);
2154 #endif
2155 	case KERN_SYSVIPC_SEM_INFO:
2156 #ifdef SYSVSEM
2157 		infosize = sizeof(semsi->seminfo);
2158 		nds = seminfo.semmni;
2159 		dssize = sizeof(semsi->semids[0]);
2160 		break;
2161 #else
2162 		return (EOPNOTSUPP);
2163 #endif
2164 	case KERN_SYSVIPC_SHM_INFO:
2165 #ifdef SYSVSHM
2166 		infosize = sizeof(shmsi->shminfo);
2167 		nds = shminfo.shmmni;
2168 		dssize = sizeof(shmsi->shmids[0]);
2169 		break;
2170 #else
2171 		return (EOPNOTSUPP);
2172 #endif
2173 	default:
2174 		return (EINVAL);
2175 	}
2176 	tsize = infosize + (nds * dssize);
2177 
2178 	/* Return just the total size required. */
2179 	if (where == NULL) {
2180 		*sizep = tsize;
2181 		return (0);
2182 	}
2183 
2184 	/* Not enough room for even the info struct. */
2185 	if (buflen < infosize) {
2186 		*sizep = 0;
2187 		return (ENOMEM);
2188 	}
2189 	bufsiz = min(tsize, buflen);
2190 	buf = malloc(bufsiz, M_TEMP, M_WAITOK|M_ZERO);
2191 
2192 	switch (*name) {
2193 #ifdef SYSVSEM
2194 	case KERN_SYSVIPC_SEM_INFO:
2195 		semsi = (struct sem_sysctl_info *)buf;
2196 		semsi->seminfo = seminfo;
2197 		break;
2198 #endif
2199 #ifdef SYSVSHM
2200 	case KERN_SYSVIPC_SHM_INFO:
2201 		shmsi = (struct shm_sysctl_info *)buf;
2202 		shmsi->shminfo = shminfo;
2203 		break;
2204 #endif
2205 	}
2206 	buflen -= infosize;
2207 
2208 	ret = 0;
2209 	if (buflen > 0) {
2210 		/* Fill in the IPC data structures.  */
2211 		for (i = 0; i < nds; i++) {
2212 			if (buflen < dssize) {
2213 				ret = ENOMEM;
2214 				break;
2215 			}
2216 			switch (*name) {
2217 #ifdef SYSVSEM
2218 			case KERN_SYSVIPC_SEM_INFO:
2219 				if (sema[i] != NULL)
2220 					memcpy(&semsi->semids[i], sema[i],
2221 					    dssize);
2222 				else
2223 					memset(&semsi->semids[i], 0, dssize);
2224 				break;
2225 #endif
2226 #ifdef SYSVSHM
2227 			case KERN_SYSVIPC_SHM_INFO:
2228 				if (shmsegs[i] != NULL)
2229 					memcpy(&shmsi->shmids[i], shmsegs[i],
2230 					    dssize);
2231 				else
2232 					memset(&shmsi->shmids[i], 0, dssize);
2233 				break;
2234 #endif
2235 			}
2236 			buflen -= dssize;
2237 		}
2238 	}
2239 	*sizep -= buflen;
2240 	error = copyout(buf, where, *sizep);
2241 	free(buf, M_TEMP, bufsiz);
2242 	/* If copyout succeeded, use return code set earlier. */
2243 	return (error ? error : ret);
2244 }
2245 #endif /* SYSVMSG || SYSVSEM || SYSVSHM */
2246 
2247 #ifndef	SMALL_KERNEL
2248 
2249 int
2250 sysctl_intrcnt(int *name, u_int namelen, void *oldp, size_t *oldlenp)
2251 {
2252 	return (evcount_sysctl(name, namelen, oldp, oldlenp, NULL, 0));
2253 }
2254 
2255 
2256 int
2257 sysctl_sensors(int *name, u_int namelen, void *oldp, size_t *oldlenp,
2258     void *newp, size_t newlen)
2259 {
2260 	struct ksensor *ks;
2261 	struct sensor *us;
2262 	struct ksensordev *ksd;
2263 	struct sensordev *usd;
2264 	int dev, numt, ret;
2265 	enum sensor_type type;
2266 
2267 	if (namelen != 1 && namelen != 3)
2268 		return (ENOTDIR);
2269 
2270 	dev = name[0];
2271 	if (namelen == 1) {
2272 		ret = sensordev_get(dev, &ksd);
2273 		if (ret)
2274 			return (ret);
2275 
2276 		/* Grab a copy, to clear the kernel pointers */
2277 		usd = malloc(sizeof(*usd), M_TEMP, M_WAITOK|M_ZERO);
2278 		usd->num = ksd->num;
2279 		strlcpy(usd->xname, ksd->xname, sizeof(usd->xname));
2280 		memcpy(usd->maxnumt, ksd->maxnumt, sizeof(usd->maxnumt));
2281 		usd->sensors_count = ksd->sensors_count;
2282 
2283 		ret = sysctl_rdstruct(oldp, oldlenp, newp, usd,
2284 		    sizeof(struct sensordev));
2285 
2286 		free(usd, M_TEMP, sizeof(*usd));
2287 		return (ret);
2288 	}
2289 
2290 	type = name[1];
2291 	numt = name[2];
2292 
2293 	ret = sensor_find(dev, type, numt, &ks);
2294 	if (ret)
2295 		return (ret);
2296 
2297 	/* Grab a copy, to clear the kernel pointers */
2298 	us = malloc(sizeof(*us), M_TEMP, M_WAITOK|M_ZERO);
2299 	memcpy(us->desc, ks->desc, sizeof(us->desc));
2300 	us->tv = ks->tv;
2301 	us->value = ks->value;
2302 	us->type = ks->type;
2303 	us->status = ks->status;
2304 	us->numt = ks->numt;
2305 	us->flags = ks->flags;
2306 
2307 	ret = sysctl_rdstruct(oldp, oldlenp, newp, us,
2308 	    sizeof(struct sensor));
2309 	free(us, M_TEMP, sizeof(*us));
2310 	return (ret);
2311 }
2312 
2313 #endif	/* SMALL_KERNEL */
2314 
2315 int
2316 sysctl_cptime2(int *name, u_int namelen, void *oldp, size_t *oldlenp,
2317     void *newp, size_t newlen)
2318 {
2319 	CPU_INFO_ITERATOR cii;
2320 	struct cpu_info *ci;
2321 	int found = 0;
2322 
2323 	if (namelen != 1)
2324 		return (ENOTDIR);
2325 
2326 	CPU_INFO_FOREACH(cii, ci) {
2327 		if (name[0] == CPU_INFO_UNIT(ci)) {
2328 			found = 1;
2329 			break;
2330 		}
2331 	}
2332 	if (!found)
2333 		return (ENOENT);
2334 
2335 	return (sysctl_rdstruct(oldp, oldlenp, newp,
2336 	    &ci->ci_schedstate.spc_cp_time,
2337 	    sizeof(ci->ci_schedstate.spc_cp_time)));
2338 }
2339