xref: /netbsd-src/sys/kern/kern_sysctl.c (revision d710132b4b8ce7f7cccaaf660cb16aa16b4077a0)
1 /*	$NetBSD: kern_sysctl.c,v 1.134 2003/06/23 11:02:05 martin Exp $	*/
2 
3 /*-
4  * Copyright (c) 1982, 1986, 1989, 1993
5  *	The Regents of the University of California.  All rights reserved.
6  *
7  * This code is derived from software contributed to Berkeley by
8  * Mike Karels at Berkeley Software Design, Inc.
9  *
10  * Redistribution and use in source and binary forms, with or without
11  * modification, are permitted provided that the following conditions
12  * are met:
13  * 1. Redistributions of source code must retain the above copyright
14  *    notice, this list of conditions and the following disclaimer.
15  * 2. Redistributions in binary form must reproduce the above copyright
16  *    notice, this list of conditions and the following disclaimer in the
17  *    documentation and/or other materials provided with the distribution.
18  * 3. All advertising materials mentioning features or use of this software
19  *    must display the following acknowledgement:
20  *	This product includes software developed by the University of
21  *	California, Berkeley and its contributors.
22  * 4. Neither the name of the University nor the names of its contributors
23  *    may be used to endorse or promote products derived from this software
24  *    without specific prior written permission.
25  *
26  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
27  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
28  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
29  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
30  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
31  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
32  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
33  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
34  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
35  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
36  * SUCH DAMAGE.
37  *
38  *	@(#)kern_sysctl.c	8.9 (Berkeley) 5/20/95
39  */
40 
41 /*
42  * sysctl system call.
43  */
44 
45 #include <sys/cdefs.h>
46 __KERNEL_RCSID(0, "$NetBSD: kern_sysctl.c,v 1.134 2003/06/23 11:02:05 martin Exp $");
47 
48 #include "opt_ddb.h"
49 #include "opt_insecure.h"
50 #include "opt_defcorename.h"
51 #include "opt_multiprocessor.h"
52 #include "opt_pipe.h"
53 #include "opt_sysv.h"
54 #include "pty.h"
55 #include "rnd.h"
56 
57 #include <sys/param.h>
58 #include <sys/systm.h>
59 #include <sys/kernel.h>
60 #include <sys/buf.h>
61 #include <sys/device.h>
62 #include <sys/disklabel.h>
63 #include <sys/dkstat.h>
64 #include <sys/exec.h>
65 #include <sys/file.h>
66 #include <sys/ioctl.h>
67 #include <sys/malloc.h>
68 #include <sys/mount.h>
69 #include <sys/msgbuf.h>
70 #include <sys/pool.h>
71 #include <sys/proc.h>
72 #include <sys/resource.h>
73 #include <sys/resourcevar.h>
74 #include <sys/sa.h>
75 #include <sys/syscallargs.h>
76 #include <sys/tty.h>
77 #include <sys/unistd.h>
78 #include <sys/vnode.h>
79 #include <sys/socketvar.h>
80 #define	__SYSCTL_PRIVATE
81 #include <sys/sysctl.h>
82 #include <sys/lock.h>
83 #include <sys/namei.h>
84 
85 #if defined(SYSVMSG) || defined(SYSVSEM) || defined(SYSVSHM)
86 #include <sys/ipc.h>
87 #endif
88 #ifdef SYSVMSG
89 #include <sys/msg.h>
90 #endif
91 #ifdef SYSVSEM
92 #include <sys/sem.h>
93 #endif
94 #ifdef SYSVSHM
95 #include <sys/shm.h>
96 #endif
97 
98 #include <dev/cons.h>
99 
100 #if defined(DDB)
101 #include <ddb/ddbvar.h>
102 #endif
103 
104 #ifndef PIPE_SOCKETPAIR
105 #include <sys/pipe.h>
106 #endif
107 
108 #if NRND > 0
109 #include <sys/rnd.h>
110 #endif
111 
112 #define PTRTOINT64(foo)	((u_int64_t)(uintptr_t)(foo))
113 
114 static int sysctl_file(void *, size_t *);
115 #if defined(SYSVMSG) || defined(SYSVSEM) || defined(SYSVSHM)
116 static int sysctl_sysvipc(int *, u_int, void *, size_t *);
117 #endif
118 static int sysctl_msgbuf(void *, size_t *);
119 static int sysctl_doeproc(int *, u_int, void *, size_t *);
120 static int sysctl_dolwp(int *, u_int, void *, size_t *);
121 static int sysctl_dotkstat(int *, u_int, void *, size_t *, void *);
122 #ifdef MULTIPROCESSOR
123 static int sysctl_docptime(void *, size_t *, void *);
124 static int sysctl_ncpus(void);
125 #endif
126 static void fill_kproc2(struct proc *, struct kinfo_proc2 *);
127 static void fill_lwp(struct lwp *, struct kinfo_lwp *);
128 static int sysctl_procargs(int *, u_int, void *, size_t *, struct proc *);
129 #if NPTY > 0
130 static int sysctl_pty(void *, size_t *, void *, size_t);
131 #endif
132 
133 /*
134  * The `sysctl_memlock' is intended to keep too many processes from
135  * locking down memory by doing sysctls at once.  Whether or not this
136  * is really a good idea to worry about it probably a subject of some
137  * debate.
138  */
139 struct lock sysctl_memlock;
140 
141 void
142 sysctl_init(void)
143 {
144 
145 	lockinit(&sysctl_memlock, PRIBIO|PCATCH, "sysctl", 0, 0);
146 }
147 
148 int
149 sys___sysctl(struct lwp *l, void *v, register_t *retval)
150 {
151 	struct sys___sysctl_args /* {
152 		syscallarg(int *) name;
153 		syscallarg(u_int) namelen;
154 		syscallarg(void *) old;
155 		syscallarg(size_t *) oldlenp;
156 		syscallarg(void *) new;
157 		syscallarg(size_t) newlen;
158 	} */ *uap = v;
159 	struct proc *p = l->l_proc;
160 	int error;
161 	size_t savelen = 0, oldlen = 0;
162 	sysctlfn *fn;
163 	int name[CTL_MAXNAME];
164 	size_t *oldlenp;
165 
166 	/*
167 	 * all top-level sysctl names are non-terminal
168 	 */
169 	if (SCARG(uap, namelen) > CTL_MAXNAME || SCARG(uap, namelen) < 2)
170 		return (EINVAL);
171 	error = copyin(SCARG(uap, name), &name,
172 	    SCARG(uap, namelen) * sizeof(int));
173 	if (error)
174 		return (error);
175 
176 	/*
177 	 * For all but CTL_PROC, must be root to change a value.
178 	 * For CTL_PROC, must be root, or owner of the proc (and not suid),
179 	 * this is checked in proc_sysctl() (once we know the targer proc).
180 	 */
181 	if (SCARG(uap, new) != NULL && name[0] != CTL_PROC &&
182 	    (error = suser(p->p_ucred, &p->p_acflag)))
183 		return (error);
184 
185 	switch (name[0]) {
186 	case CTL_KERN:
187 		fn = kern_sysctl;
188 		break;
189 	case CTL_HW:
190 		fn = hw_sysctl;
191 		break;
192 	case CTL_VM:
193 		fn = uvm_sysctl;
194 		break;
195 	case CTL_NET:
196 		fn = net_sysctl;
197 		break;
198 	case CTL_VFS:
199 		fn = vfs_sysctl;
200 		break;
201 	case CTL_MACHDEP:
202 		fn = cpu_sysctl;
203 		break;
204 #ifdef DEBUG
205 	case CTL_DEBUG:
206 		fn = debug_sysctl;
207 		break;
208 #endif
209 #ifdef DDB
210 	case CTL_DDB:
211 		fn = ddb_sysctl;
212 		break;
213 #endif
214 	case CTL_PROC:
215 		fn = proc_sysctl;
216 		break;
217 
218 	case CTL_EMUL:
219 		fn = emul_sysctl;
220 		break;
221 	default:
222 		return (EOPNOTSUPP);
223 	}
224 
225 	/*
226 	 * XXX Hey, we wire `old', but what about `new'?
227 	 */
228 
229 	oldlenp = SCARG(uap, oldlenp);
230 	if (oldlenp) {
231 		if ((error = copyin(oldlenp, &oldlen, sizeof(oldlen))))
232 			return (error);
233 		oldlenp = &oldlen;
234 	}
235 	if (SCARG(uap, old) != NULL) {
236 		error = lockmgr(&sysctl_memlock, LK_EXCLUSIVE, NULL);
237 		if (error)
238 			return (error);
239 		error = uvm_vslock(p, SCARG(uap, old), oldlen, VM_PROT_WRITE);
240 		if (error) {
241 			(void) lockmgr(&sysctl_memlock, LK_RELEASE, NULL);
242 			return (error);
243 		}
244 		savelen = oldlen;
245 	}
246 	error = (*fn)(name + 1, SCARG(uap, namelen) - 1, SCARG(uap, old),
247 	    oldlenp, SCARG(uap, new), SCARG(uap, newlen), p);
248 	if (SCARG(uap, old) != NULL) {
249 		uvm_vsunlock(p, SCARG(uap, old), savelen);
250 		(void) lockmgr(&sysctl_memlock, LK_RELEASE, NULL);
251 	}
252 	if (error)
253 		return (error);
254 	if (SCARG(uap, oldlenp))
255 		error = copyout(&oldlen, SCARG(uap, oldlenp), sizeof(oldlen));
256 	return (error);
257 }
258 
259 /*
260  * Attributes stored in the kernel.
261  */
262 char hostname[MAXHOSTNAMELEN];
263 int hostnamelen;
264 
265 char domainname[MAXHOSTNAMELEN];
266 int domainnamelen;
267 
268 long hostid;
269 
270 #ifdef INSECURE
271 int securelevel = -1;
272 #else
273 int securelevel = 0;
274 #endif
275 
276 #ifndef DEFCORENAME
277 #define	DEFCORENAME	"%n.core"
278 #endif
279 char defcorename[MAXPATHLEN] = DEFCORENAME;
280 int defcorenamelen = sizeof(DEFCORENAME);
281 
282 extern	int	kern_logsigexit;
283 extern	fixpt_t	ccpu;
284 extern	int	forkfsleep;
285 extern	int	dumponpanic;
286 
287 #ifndef MULTIPROCESSOR
288 #define sysctl_ncpus() 1
289 #endif
290 
291 #ifdef MULTIPROCESSOR
292 
293 #ifndef CPU_INFO_FOREACH
294 #define CPU_INFO_ITERATOR int
295 #define CPU_INFO_FOREACH(cii, ci) cii = 0, ci = curcpu(); ci != NULL; ci = NULL
296 #endif
297 
298 static int
299 sysctl_docptime(void *oldp, size_t *oldlenp, void *newp)
300 {
301 	u_int64_t cp_time[CPUSTATES];
302 	int i;
303 	struct cpu_info *ci;
304 	CPU_INFO_ITERATOR cii;
305 
306 	for (i = 0; i < CPUSTATES; i++)
307 		cp_time[i] = 0;
308 
309 	for (CPU_INFO_FOREACH(cii, ci)) {
310 		for (i = 0; i < CPUSTATES; i++)
311 			cp_time[i] += ci->ci_schedstate.spc_cp_time[i];
312 	}
313 	return (sysctl_rdstruct(oldp, oldlenp, newp,
314 	    cp_time, sizeof(cp_time)));
315 }
316 
317 static int
318 sysctl_ncpus(void)
319 {
320 	struct cpu_info *ci;
321 	CPU_INFO_ITERATOR cii;
322 
323 	int ncpus = 0;
324 	for (CPU_INFO_FOREACH(cii, ci))
325 		ncpus++;
326 	return (ncpus);
327 }
328 
329 #endif
330 
331 /*
332  * kernel related system variables.
333  */
334 int
335 kern_sysctl(int *name, u_int namelen, void *oldp, size_t *oldlenp,
336     void *newp, size_t newlen, struct proc *p)
337 {
338 	int error, level, inthostid;
339 	int old_autonicetime;
340 	int old_vnodes;
341 	dev_t consdev;
342 #if NRND > 0
343 	int v;
344 #endif
345 
346 	/* All sysctl names at this level, except for a few, are terminal. */
347 	switch (name[0]) {
348 	case KERN_PROC:
349 	case KERN_PROC2:
350 	case KERN_LWP:
351 	case KERN_PROF:
352 	case KERN_MBUF:
353 	case KERN_PROC_ARGS:
354 	case KERN_SYSVIPC_INFO:
355 	case KERN_PIPE:
356 	case KERN_TKSTAT:
357 		/* Not terminal. */
358 		break;
359 	default:
360 		if (namelen != 1)
361 			return (ENOTDIR);	/* overloaded */
362 	}
363 
364 	switch (name[0]) {
365 	case KERN_OSTYPE:
366 		return (sysctl_rdstring(oldp, oldlenp, newp, ostype));
367 	case KERN_OSRELEASE:
368 		return (sysctl_rdstring(oldp, oldlenp, newp, osrelease));
369 	case KERN_OSREV:
370 		return (sysctl_rdint(oldp, oldlenp, newp, __NetBSD_Version__));
371 	case KERN_VERSION:
372 		return (sysctl_rdstring(oldp, oldlenp, newp, version));
373 	case KERN_MAXVNODES:
374 		old_vnodes = desiredvnodes;
375 		error = sysctl_int(oldp, oldlenp, newp, newlen, &desiredvnodes);
376 		if (newp && !error) {
377 			if (old_vnodes > desiredvnodes) {
378 				desiredvnodes = old_vnodes;
379 				return (EINVAL);
380 			}
381 			vfs_reinit();
382 			nchreinit();
383 		}
384 		return (error);
385 	case KERN_MAXPROC:
386 	    {
387 		int nmaxproc = maxproc;
388 
389 		error = sysctl_int(oldp, oldlenp, newp, newlen, &nmaxproc);
390 
391 		if (!error && newp) {
392 			if (nmaxproc < 0 || nmaxproc >= PID_MAX)
393 				return (EINVAL);
394 
395 #ifdef __HAVE_CPU_MAXPROC
396 			if (nmaxproc > cpu_maxproc())
397 				return (EINVAL);
398 #endif
399 			maxproc = nmaxproc;
400 		}
401 
402 		return (error);
403 	    }
404 	case KERN_MAXFILES:
405 		return (sysctl_int(oldp, oldlenp, newp, newlen, &maxfiles));
406 	case KERN_ARGMAX:
407 		return (sysctl_rdint(oldp, oldlenp, newp, ARG_MAX));
408 	case KERN_SECURELVL:
409 		level = securelevel;
410 		if ((error = sysctl_int(oldp, oldlenp, newp, newlen, &level)) ||
411 		    newp == NULL)
412 			return (error);
413 		if (level < securelevel && p->p_pid != 1)
414 			return (EPERM);
415 		securelevel = level;
416 		return (0);
417 	case KERN_HOSTNAME:
418 		error = sysctl_string(oldp, oldlenp, newp, newlen,
419 		    hostname, sizeof(hostname));
420 		if (newp && !error)
421 			hostnamelen = newlen;
422 		return (error);
423 	case KERN_DOMAINNAME:
424 		error = sysctl_string(oldp, oldlenp, newp, newlen,
425 		    domainname, sizeof(domainname));
426 		if (newp && !error)
427 			domainnamelen = newlen;
428 		return (error);
429 	case KERN_HOSTID:
430 		inthostid = hostid;  /* XXX assumes sizeof long <= sizeof int */
431 		error = sysctl_int(oldp, oldlenp, newp, newlen, &inthostid);
432 		if (newp && !error)
433 			hostid = inthostid;
434 		return (error);
435 	case KERN_CLOCKRATE:
436 		return (sysctl_clockrate(oldp, oldlenp));
437 	case KERN_BOOTTIME:
438 		return (sysctl_rdstruct(oldp, oldlenp, newp, &boottime,
439 		    sizeof(struct timeval)));
440 	case KERN_VNODE:
441 		return (sysctl_vnode(oldp, oldlenp, p));
442 	case KERN_PROC:
443 	case KERN_PROC2:
444 		return (sysctl_doeproc(name, namelen, oldp, oldlenp));
445 	case KERN_LWP:
446 		return (sysctl_dolwp(name, namelen, oldp, oldlenp));
447 	case KERN_PROC_ARGS:
448 		return (sysctl_procargs(name + 1, namelen - 1,
449 		    oldp, oldlenp, p));
450 	case KERN_FILE:
451 		return (sysctl_file(oldp, oldlenp));
452 #ifdef GPROF
453 	case KERN_PROF:
454 		return (sysctl_doprof(name + 1, namelen - 1, oldp, oldlenp,
455 		    newp, newlen));
456 #endif
457 	case KERN_POSIX1:
458 		return (sysctl_rdint(oldp, oldlenp, newp, _POSIX_VERSION));
459 	case KERN_NGROUPS:
460 		return (sysctl_rdint(oldp, oldlenp, newp, NGROUPS_MAX));
461 	case KERN_JOB_CONTROL:
462 		return (sysctl_rdint(oldp, oldlenp, newp, 1));
463 	case KERN_SAVED_IDS:
464 #ifdef _POSIX_SAVED_IDS
465 		return (sysctl_rdint(oldp, oldlenp, newp, 1));
466 #else
467 		return (sysctl_rdint(oldp, oldlenp, newp, 0));
468 #endif
469 	case KERN_MAXPARTITIONS:
470 		return (sysctl_rdint(oldp, oldlenp, newp, MAXPARTITIONS));
471 	case KERN_RAWPARTITION:
472 		return (sysctl_rdint(oldp, oldlenp, newp, RAW_PART));
473 #ifdef NTP
474 	case KERN_NTPTIME:
475 		return (sysctl_ntptime(oldp, oldlenp));
476 #endif
477 	case KERN_AUTONICETIME:
478 		old_autonicetime = autonicetime;
479 		error = sysctl_int(oldp, oldlenp, newp, newlen, &autonicetime);
480 		if (autonicetime < 0)
481  			autonicetime = old_autonicetime;
482 		return (error);
483 	case KERN_AUTONICEVAL:
484 		error = sysctl_int(oldp, oldlenp, newp, newlen, &autoniceval);
485 		if (autoniceval < PRIO_MIN)
486 			autoniceval = PRIO_MIN;
487 		if (autoniceval > PRIO_MAX)
488 			autoniceval = PRIO_MAX;
489 		return (error);
490 	case KERN_RTC_OFFSET:
491 		return (sysctl_rdint(oldp, oldlenp, newp, rtc_offset));
492 	case KERN_ROOT_DEVICE:
493 		return (sysctl_rdstring(oldp, oldlenp, newp,
494 		    root_device->dv_xname));
495 	case KERN_MSGBUFSIZE:
496 		/*
497 		 * deal with cases where the message buffer has
498 		 * become corrupted.
499 		 */
500 		if (!msgbufenabled || msgbufp->msg_magic != MSG_MAGIC) {
501 			msgbufenabled = 0;
502 			return (ENXIO);
503 		}
504 		return (sysctl_rdint(oldp, oldlenp, newp, msgbufp->msg_bufs));
505 	case KERN_FSYNC:
506 		return (sysctl_rdint(oldp, oldlenp, newp, 1));
507 	case KERN_SYSVMSG:
508 #ifdef SYSVMSG
509 		return (sysctl_rdint(oldp, oldlenp, newp, 1));
510 #else
511 		return (sysctl_rdint(oldp, oldlenp, newp, 0));
512 #endif
513 	case KERN_SYSVSEM:
514 #ifdef SYSVSEM
515 		return (sysctl_rdint(oldp, oldlenp, newp, 1));
516 #else
517 		return (sysctl_rdint(oldp, oldlenp, newp, 0));
518 #endif
519 	case KERN_SYSVSHM:
520 #ifdef SYSVSHM
521 		return (sysctl_rdint(oldp, oldlenp, newp, 1));
522 #else
523 		return (sysctl_rdint(oldp, oldlenp, newp, 0));
524 #endif
525  	case KERN_DEFCORENAME:
526 		if (newp && newlen < 1)
527 			return (EINVAL);
528 		error = sysctl_string(oldp, oldlenp, newp, newlen,
529 		    defcorename, sizeof(defcorename));
530 		if (newp && !error)
531 			defcorenamelen = newlen;
532 		return (error);
533 	case KERN_SYNCHRONIZED_IO:
534 		return (sysctl_rdint(oldp, oldlenp, newp, 1));
535 	case KERN_IOV_MAX:
536 		return (sysctl_rdint(oldp, oldlenp, newp, IOV_MAX));
537 	case KERN_MBUF:
538 		return (sysctl_dombuf(name + 1, namelen - 1, oldp, oldlenp,
539 		    newp, newlen));
540 	case KERN_MAPPED_FILES:
541 		return (sysctl_rdint(oldp, oldlenp, newp, 1));
542 	case KERN_MEMLOCK:
543 		return (sysctl_rdint(oldp, oldlenp, newp, 1));
544 	case KERN_MEMLOCK_RANGE:
545 		return (sysctl_rdint(oldp, oldlenp, newp, 1));
546 	case KERN_MEMORY_PROTECTION:
547 		return (sysctl_rdint(oldp, oldlenp, newp, 1));
548 	case KERN_LOGIN_NAME_MAX:
549 		return (sysctl_rdint(oldp, oldlenp, newp, LOGIN_NAME_MAX));
550 	case KERN_LOGSIGEXIT:
551 		return (sysctl_int(oldp, oldlenp, newp, newlen,
552 		    &kern_logsigexit));
553 	case KERN_FSCALE:
554 		return (sysctl_rdint(oldp, oldlenp, newp, FSCALE));
555 	case KERN_CCPU:
556 		return (sysctl_rdint(oldp, oldlenp, newp, ccpu));
557 	case KERN_CP_TIME:
558 #ifndef MULTIPROCESSOR
559 		return (sysctl_rdstruct(oldp, oldlenp, newp,
560 		    curcpu()->ci_schedstate.spc_cp_time,
561 		    sizeof(curcpu()->ci_schedstate.spc_cp_time)));
562 #else
563 		return (sysctl_docptime(oldp, oldlenp, newp));
564 #endif
565 #if defined(SYSVMSG) || defined(SYSVSEM) || defined(SYSVSHM)
566 	case KERN_SYSVIPC_INFO:
567 		return (sysctl_sysvipc(name + 1, namelen - 1, oldp, oldlenp));
568 #endif
569 	case KERN_MSGBUF:
570 		return (sysctl_msgbuf(oldp, oldlenp));
571 	case KERN_CONSDEV:
572 		if (cn_tab != NULL)
573 			consdev = cn_tab->cn_dev;
574 		else
575 			consdev = NODEV;
576 		return (sysctl_rdstruct(oldp, oldlenp, newp, &consdev,
577 		    sizeof consdev));
578 #if NPTY > 0
579 	case KERN_MAXPTYS:
580 		return (sysctl_pty(oldp, oldlenp, newp, newlen));
581 #endif
582 #ifndef PIPE_SOCKETPAIR
583 	case KERN_PIPE:
584 		return (sysctl_dopipe(name + 1, namelen - 1, oldp, oldlenp,
585 		    newp, newlen));
586 #endif
587 	case KERN_MAXPHYS:
588 		return (sysctl_rdint(oldp, oldlenp, newp, MAXPHYS));
589 	case KERN_SBMAX:
590 	    {
591 		int new_sbmax = sb_max;
592 
593 		error = sysctl_int(oldp, oldlenp, newp, newlen, &new_sbmax);
594 		if (newp && !error) {
595 			if (new_sbmax < (16 * 1024)) /* sanity */
596 				return (EINVAL);
597 			sb_max = new_sbmax;
598 		}
599 		return (error);
600 	    }
601 	case KERN_TKSTAT:
602 		return (sysctl_dotkstat(name + 1, namelen - 1, oldp, oldlenp,
603 		    newp));
604 	case KERN_MONOTONIC_CLOCK:	/* XXX _POSIX_VERSION */
605 		return (sysctl_rdint(oldp, oldlenp, newp, 200112));
606 	case KERN_URND:
607 #if NRND > 0
608 		if (rnd_extract_data(&v, sizeof(v), RND_EXTRACT_ANY) ==
609 		    sizeof(v))
610 			return (sysctl_rdint(oldp, oldlenp, newp, v));
611 		else
612 			return (EIO);	/*XXX*/
613 #else
614 		return (EOPNOTSUPP);
615 #endif
616 	case KERN_LABELSECTOR:
617 		return (sysctl_rdint(oldp, oldlenp, newp, LABELSECTOR));
618 	case KERN_LABELOFFSET:
619 		return (sysctl_rdint(oldp, oldlenp, newp, LABELOFFSET));
620 	case KERN_FORKFSLEEP:
621 	    {
622 		/* userland sees value in ms, internally is in ticks */
623 		int timo, lsleep = forkfsleep * 1000 / hz;
624 
625 		error = sysctl_int(oldp, oldlenp, newp, newlen, &lsleep);
626 		if (newp && !error) {
627 			/* refuse negative values, and overly 'long time' */
628 			if (lsleep < 0 || lsleep > MAXSLP * 1000)
629 				return (EINVAL);
630 
631 			timo = mstohz(lsleep);
632 
633 			/* if the interval is >0 ms && <1 tick, use 1 tick */
634 			if (lsleep != 0 && timo == 0)
635 				forkfsleep = 1;
636 			else
637 				forkfsleep = timo;
638 		}
639 		return (error);
640 	    }
641 	case KERN_POSIX_THREADS:	/* XXX _POSIX_VERSION */
642 		return (sysctl_rdint(oldp, oldlenp, newp, 200112));
643 	case KERN_POSIX_SEMAPHORES:	/* XXX _POSIX_VERSION */
644 #ifdef P1003_1B_SEMAPHORE
645 		return (sysctl_rdint(oldp, oldlenp, newp, 200112));
646 #else
647 		return (sysctl_rdint(oldp, oldlenp, newp, 0));
648 #endif
649 	case KERN_POSIX_BARRIERS:	/* XXX _POSIX_VERSION */
650 		return (sysctl_rdint(oldp, oldlenp, newp, 200112));
651 	case KERN_POSIX_TIMERS:		/* XXX _POSIX_VERSION */
652 		return (sysctl_rdint(oldp, oldlenp, newp, 200112));
653 	case KERN_POSIX_SPIN_LOCKS:	/* XXX _POSIX_VERSION */
654 		return (sysctl_rdint(oldp, oldlenp, newp, 200112));
655 	case KERN_POSIX_READER_WRITER_LOCKS:	/* XXX _POSIX_VERSION */
656 		return (sysctl_rdint(oldp, oldlenp, newp, 200112));
657 	case KERN_DUMP_ON_PANIC:
658 		return (sysctl_int(oldp, oldlenp, newp, newlen, &dumponpanic));
659 
660 	default:
661 		return (EOPNOTSUPP);
662 	}
663 	/* NOTREACHED */
664 }
665 
666 /*
667  * hardware related system variables.
668  */
669 int
670 hw_sysctl(int *name, u_int namelen, void *oldp, size_t *oldlenp,
671     void *newp, size_t newlen, struct proc *p)
672 {
673 
674 	/* All sysctl names at this level, except for a few, are terminal. */
675 	switch (name[0]) {
676 	case HW_DISKSTATS:
677 		/* Not terminal. */
678 		break;
679 	default:
680 		if (namelen != 1)
681 			return (ENOTDIR);	/* overloaded */
682 	}
683 
684 	switch (name[0]) {
685 	case HW_MACHINE:
686 		return (sysctl_rdstring(oldp, oldlenp, newp, machine));
687 	case HW_MACHINE_ARCH:
688 		return (sysctl_rdstring(oldp, oldlenp, newp, machine_arch));
689 	case HW_MODEL:
690 		return (sysctl_rdstring(oldp, oldlenp, newp, cpu_model));
691 	case HW_NCPU:
692 		return (sysctl_rdint(oldp, oldlenp, newp, sysctl_ncpus()));
693 	case HW_BYTEORDER:
694 		return (sysctl_rdint(oldp, oldlenp, newp, BYTE_ORDER));
695 	case HW_PHYSMEM:
696 	    {
697 		u_int rval;
698 
699 		if ((u_int)physmem > (UINT_MAX / PAGE_SIZE))
700 			rval = UINT_MAX;
701 		else
702 			rval = physmem * PAGE_SIZE;
703 		return (sysctl_rdint(oldp, oldlenp, newp, rval));
704 	    }
705 	case HW_PHYSMEM64:
706 		return (sysctl_rdquad(oldp, oldlenp, newp,
707 		    (u_quad_t)physmem * PAGE_SIZE));
708 	case HW_USERMEM:
709 	    {
710 		u_int rval;
711 
712 		if ((u_int)(physmem - uvmexp.wired) > (UINT_MAX / PAGE_SIZE))
713 			rval = UINT_MAX;
714 		else
715 			rval = (physmem - uvmexp.wired) * PAGE_SIZE;
716 		return (sysctl_rdint(oldp, oldlenp, newp, rval));
717 	    }
718 	case HW_USERMEM64:
719 		return (sysctl_rdquad(oldp, oldlenp, newp,
720 		    (u_quad_t)(physmem - uvmexp.wired) * PAGE_SIZE));
721 	case HW_PAGESIZE:
722 		return (sysctl_rdint(oldp, oldlenp, newp, PAGE_SIZE));
723 	case HW_ALIGNBYTES:
724 		return (sysctl_rdint(oldp, oldlenp, newp, ALIGNBYTES));
725 	case HW_DISKNAMES:
726 		return (sysctl_disknames(oldp, oldlenp));
727 	case HW_DISKSTATS:
728 		return (sysctl_diskstats(name + 1, namelen - 1, oldp, oldlenp));
729 	case HW_CNMAGIC: {
730 		char magic[CNS_LEN];
731 		int error;
732 
733 		if (oldp)
734 			cn_get_magic(magic, CNS_LEN);
735 		error = sysctl_string(oldp, oldlenp, newp, newlen,
736 		    magic, sizeof(magic));
737 		if (newp && !error) {
738 			error = cn_set_magic(magic);
739 		}
740 		return (error);
741 	}
742 	default:
743 		return (EOPNOTSUPP);
744 	}
745 	/* NOTREACHED */
746 }
747 
748 #ifdef DEBUG
749 /*
750  * Debugging related system variables.
751  */
752 struct ctldebug /* debug0, */ /* debug1, */ debug2, debug3, debug4;
753 struct ctldebug debug5, debug6, debug7, debug8, debug9;
754 struct ctldebug debug10, debug11, debug12, debug13, debug14;
755 struct ctldebug debug15, debug16, debug17, debug18, debug19;
756 static struct ctldebug *debugvars[CTL_DEBUG_MAXID] = {
757 	&debug0, &debug1, &debug2, &debug3, &debug4,
758 	&debug5, &debug6, &debug7, &debug8, &debug9,
759 	&debug10, &debug11, &debug12, &debug13, &debug14,
760 	&debug15, &debug16, &debug17, &debug18, &debug19,
761 };
762 
763 int
764 debug_sysctl(int *name, u_int namelen, void *oldp, size_t *oldlenp,
765     void *newp, size_t newlen, struct proc *p)
766 {
767 	struct ctldebug *cdp;
768 
769 	/* all sysctl names at this level are name and field */
770 	if (namelen != 2)
771 		return (ENOTDIR);		/* overloaded */
772 	if (name[0] >= CTL_DEBUG_MAXID)
773 		return (EOPNOTSUPP);
774 	cdp = debugvars[name[0]];
775 	if (cdp->debugname == 0)
776 		return (EOPNOTSUPP);
777 	switch (name[1]) {
778 	case CTL_DEBUG_NAME:
779 		return (sysctl_rdstring(oldp, oldlenp, newp, cdp->debugname));
780 	case CTL_DEBUG_VALUE:
781 		return (sysctl_int(oldp, oldlenp, newp, newlen, cdp->debugvar));
782 	default:
783 		return (EOPNOTSUPP);
784 	}
785 	/* NOTREACHED */
786 }
787 #endif /* DEBUG */
788 
789 int
790 proc_sysctl(int *name, u_int namelen, void *oldp, size_t *oldlenp,
791     void *newp, size_t newlen, struct proc *p)
792 {
793 	struct proc *ptmp = NULL;
794 	const struct proclist_desc *pd;
795 	int error = 0;
796 	struct rlimit alim;
797 	struct plimit *newplim;
798 	char *tmps = NULL;
799 	size_t len, curlen;
800 	u_int i;
801 
802 	if (namelen < 2)
803 		return (EINVAL);
804 
805 	if (name[0] == PROC_CURPROC) {
806 		ptmp = p;
807 	} else {
808 		proclist_lock_read();
809 		for (pd = proclists; pd->pd_list != NULL; pd++) {
810 			for (ptmp = LIST_FIRST(pd->pd_list); ptmp != NULL;
811 			    ptmp = LIST_NEXT(ptmp, p_list)) {
812 				/* Skip embryonic processes. */
813 				if (ptmp->p_stat == SIDL)
814 					continue;
815 				if (ptmp->p_pid == (pid_t)name[0])
816 					break;
817 			}
818 			if (ptmp != NULL)
819 				break;
820 		}
821 		proclist_unlock_read();
822 		if (ptmp == NULL)
823 			return (ESRCH);
824 		if (p->p_ucred->cr_uid != 0) {
825 			if (p->p_cred->p_ruid != ptmp->p_cred->p_ruid ||
826 			    p->p_cred->p_ruid != ptmp->p_cred->p_svuid)
827 				return (EPERM);
828 			if (ptmp->p_cred->p_rgid != ptmp->p_cred->p_svgid)
829 				return (EPERM); /* sgid proc */
830 			for (i = 0; i < p->p_ucred->cr_ngroups; i++) {
831 				if (p->p_ucred->cr_groups[i] ==
832 				    ptmp->p_cred->p_rgid)
833 					break;
834 			}
835 			if (i == p->p_ucred->cr_ngroups)
836 				return (EPERM);
837 		}
838 	}
839 	switch (name[1]) {
840 	case PROC_PID_STOPFORK:
841 		if (namelen != 2)
842 			return (EINVAL);
843 		i = ((ptmp->p_flag & P_STOPFORK) != 0);
844 		if ((error = sysctl_int(oldp, oldlenp, newp, newlen, &i)) != 0)
845 			return (error);
846 		if (i != 0)
847 			ptmp->p_flag |= P_STOPFORK;
848 		else
849 			ptmp->p_flag &= ~P_STOPFORK;
850 		return (0);
851 		break;
852 
853 	case PROC_PID_STOPEXEC:
854 		if (namelen != 2)
855 			return (EINVAL);
856 		i = ((ptmp->p_flag & P_STOPEXEC) != 0);
857 		if ((error = sysctl_int(oldp, oldlenp, newp, newlen, &i)) != 0)
858 			return (error);
859 		if (i != 0)
860 			ptmp->p_flag |= P_STOPEXEC;
861 		else
862 			ptmp->p_flag &= ~P_STOPEXEC;
863 		return (0);
864 		break;
865 
866 	case PROC_PID_CORENAME:
867 		if (namelen != 2)
868 			return (EINVAL);
869 		/*
870 		 * Can't use sysctl_string() here because we may malloc a new
871 		 * area during the process, so we have to do it by hand.
872 		 */
873 		curlen = strlen(ptmp->p_limit->pl_corename) + 1;
874 		if (oldlenp && *oldlenp < curlen) {
875 			if (!oldp)
876 				*oldlenp = curlen;
877 			return (ENOMEM);
878 		}
879 		if (newp) {
880 			if (securelevel > 2)
881 				return (EPERM);
882 			if (newlen > MAXPATHLEN)
883 				return (ENAMETOOLONG);
884 			tmps = malloc(newlen + 1, M_TEMP, M_WAITOK);
885 			if (tmps == NULL)
886 				return (ENOMEM);
887 			error = copyin(newp, tmps, newlen + 1);
888 			tmps[newlen] = '\0';
889 			if (error)
890 				goto cleanup;
891 			/* Enforce to be either 'core' for end with '.core' */
892 			if (newlen < 4) {	/* c.o.r.e */
893 				error = EINVAL;
894 				goto cleanup;
895 			}
896 			len = newlen - 4;
897 			if (len > 0) {
898 				if (tmps[len - 1] != '.' &&
899 				    tmps[len - 1] != '/') {
900 					error = EINVAL;
901 					goto cleanup;
902 				}
903 			}
904 			if (strcmp(&tmps[len], "core") != 0) {
905 				error = EINVAL;
906 				goto cleanup;
907 			}
908 		}
909 		if (oldp && oldlenp) {
910 			*oldlenp = curlen;
911 			error = copyout(ptmp->p_limit->pl_corename, oldp,
912 			    curlen);
913 		}
914 		if (newp && error == 0) {
915 			/* if the 2 strings are identical, don't limcopy() */
916 			if (strcmp(tmps, ptmp->p_limit->pl_corename) == 0) {
917 				error = 0;
918 				goto cleanup;
919 			}
920 			if (ptmp->p_limit->p_refcnt > 1 &&
921 			    (ptmp->p_limit->p_lflags & PL_SHAREMOD) == 0) {
922 				newplim = limcopy(ptmp->p_limit);
923 				limfree(ptmp->p_limit);
924 				ptmp->p_limit = newplim;
925 			}
926 			if (ptmp->p_limit->pl_corename != defcorename) {
927 				free(ptmp->p_limit->pl_corename, M_TEMP);
928 			}
929 			ptmp->p_limit->pl_corename = tmps;
930 			return (0);
931 		}
932 cleanup:
933 		if (tmps)
934 			free(tmps, M_TEMP);
935 		return (error);
936 		break;
937 
938 	case PROC_PID_LIMIT:
939 		if (namelen != 4 || name[2] >= PROC_PID_LIMIT_MAXID)
940 			return (EINVAL);
941 		memcpy(&alim, &ptmp->p_rlimit[name[2] - 1], sizeof(alim));
942 		if (name[3] == PROC_PID_LIMIT_TYPE_HARD)
943 			error = sysctl_quad(oldp, oldlenp, newp, newlen,
944 			    &alim.rlim_max);
945 		else if (name[3] == PROC_PID_LIMIT_TYPE_SOFT)
946 			error = sysctl_quad(oldp, oldlenp, newp, newlen,
947 			    &alim.rlim_cur);
948 		else
949 			error = (EINVAL);
950 
951 		if (error)
952 			return (error);
953 
954 		if (newp)
955 			error = dosetrlimit(ptmp, p->p_cred,
956 			    name[2] - 1, &alim);
957 		return (error);
958 		break;
959 
960 	default:
961 		return (EINVAL);
962 		break;
963 	}
964 	/* NOTREACHED */
965 	return (EINVAL);
966 }
967 
968 int
969 emul_sysctl(int *name, u_int namelen, void *oldp, size_t *oldlenp,
970     void *newp, size_t newlen, struct proc *p)
971 {
972 	static struct {
973 		const char *name;
974 		int type;
975 	} emulations[] = CTL_EMUL_NAMES;
976 	const struct emul *e;
977 	const char *ename;
978 #ifdef LKM
979 	extern struct lock exec_lock;	/* XXX */
980 	int error;
981 #else
982 	extern int nexecs_builtin;
983 	extern const struct execsw execsw_builtin[];
984 	int i;
985 #endif
986 
987 	/* all sysctl names at this level are name and field */
988 	if (namelen < 2)
989 		return (ENOTDIR);		/* overloaded */
990 
991 	if ((u_int) name[0] >= EMUL_MAXID || name[0] == 0)
992 		return (EOPNOTSUPP);
993 
994 	ename = emulations[name[0]].name;
995 
996 #ifdef LKM
997 	lockmgr(&exec_lock, LK_SHARED, NULL);
998 	if ((e = emul_search(ename))) {
999 		error = (*e->e_sysctl)(name + 1, namelen - 1, oldp, oldlenp,
1000 		    newp, newlen, p);
1001 	} else
1002 		error = EOPNOTSUPP;
1003 	lockmgr(&exec_lock, LK_RELEASE, NULL);
1004 
1005 	return (error);
1006 #else
1007 	for (i = 0; i < nexecs_builtin; i++) {
1008 		e = execsw_builtin[i].es_emul;
1009 		/*
1010 		 * In order to match e.g. e->e_name "irix o32"
1011 		 * with ename "irix", we limit the comparison
1012 		 * to the length of ename.
1013 		 */
1014 		if (e == NULL ||
1015 		    strncmp(ename, e->e_name, strlen(ename)) != 0 ||
1016 		    e->e_sysctl == NULL)
1017 			continue;
1018 
1019 		return ((*e->e_sysctl)(name + 1, namelen - 1, oldp, oldlenp,
1020 		    newp, newlen, p));
1021 	}
1022 
1023 	return (EOPNOTSUPP);
1024 #endif
1025 }
1026 /*
1027  * Convenience macros.
1028  */
1029 
1030 #define SYSCTL_SCALAR_CORE_LEN(oldp, oldlenp, valp, len) 		\
1031 	if (oldlenp) {							\
1032 		if (!oldp)						\
1033 			*oldlenp = len;					\
1034 		else {							\
1035 			if (*oldlenp < len)				\
1036 				return (ENOMEM);			\
1037 			*oldlenp = len;					\
1038 			error = copyout((caddr_t)valp, oldp, len);	\
1039 		}							\
1040 	}
1041 
1042 #define SYSCTL_SCALAR_CORE_TYP(oldp, oldlenp, valp, typ) \
1043 	SYSCTL_SCALAR_CORE_LEN(oldp, oldlenp, valp, sizeof(typ))
1044 
1045 #define SYSCTL_SCALAR_NEWPCHECK_LEN(newp, newlen, len)	\
1046 	if (newp && newlen != len)			\
1047 		return (EINVAL);
1048 
1049 #define SYSCTL_SCALAR_NEWPCHECK_TYP(newp, newlen, typ)	\
1050 	SYSCTL_SCALAR_NEWPCHECK_LEN(newp, newlen, sizeof(typ))
1051 
1052 #define SYSCTL_SCALAR_NEWPCOP_LEN(newp, valp, len)	\
1053 	if (error == 0 && newp)				\
1054 		error = copyin(newp, valp, len);
1055 
1056 #define SYSCTL_SCALAR_NEWPCOP_TYP(newp, valp, typ)	\
1057 	SYSCTL_SCALAR_NEWPCOP_LEN(newp, valp, sizeof(typ))
1058 
1059 #define SYSCTL_STRING_CORE(oldp, oldlenp, str)		\
1060 	if (oldlenp) {					\
1061 		len = strlen(str) + 1;			\
1062 		if (!oldp)				\
1063 			*oldlenp = len;			\
1064 		else {					\
1065 			if (*oldlenp < len) {		\
1066 				err2 = ENOMEM;		\
1067 				len = *oldlenp;		\
1068 			} else				\
1069 				*oldlenp = len;		\
1070 			error = copyout(str, oldp, len);\
1071 			if (error == 0)			\
1072 				error = err2;		\
1073 		}					\
1074 	}
1075 
1076 /*
1077  * Validate parameters and get old / set new parameters
1078  * for an integer-valued sysctl function.
1079  */
1080 int
1081 sysctl_int(void *oldp, size_t *oldlenp, void *newp, size_t newlen, int *valp)
1082 {
1083 	int error = 0;
1084 
1085 	SYSCTL_SCALAR_NEWPCHECK_TYP(newp, newlen, int)
1086 	SYSCTL_SCALAR_CORE_TYP(oldp, oldlenp, valp, int)
1087 	SYSCTL_SCALAR_NEWPCOP_TYP(newp, valp, int)
1088 
1089 	return (error);
1090 }
1091 
1092 
1093 /*
1094  * As above, but read-only.
1095  */
1096 int
1097 sysctl_rdint(void *oldp, size_t *oldlenp, void *newp, int val)
1098 {
1099 	int error = 0;
1100 
1101 	if (newp)
1102 		return (EPERM);
1103 
1104 	SYSCTL_SCALAR_CORE_TYP(oldp, oldlenp, &val, int)
1105 
1106 	return (error);
1107 }
1108 
1109 /*
1110  * Validate parameters and get old / set new parameters
1111  * for an quad-valued sysctl function.
1112  */
1113 int
1114 sysctl_quad(void *oldp, size_t *oldlenp, void *newp, size_t newlen,
1115     quad_t *valp)
1116 {
1117 	int error = 0;
1118 
1119 	SYSCTL_SCALAR_NEWPCHECK_TYP(newp, newlen, quad_t)
1120 	SYSCTL_SCALAR_CORE_TYP(oldp, oldlenp, valp, quad_t)
1121 	SYSCTL_SCALAR_NEWPCOP_TYP(newp, valp, quad_t)
1122 
1123 	return (error);
1124 }
1125 
1126 /*
1127  * As above, but read-only.
1128  */
1129 int
1130 sysctl_rdquad(void *oldp, size_t *oldlenp, void *newp, quad_t val)
1131 {
1132 	int error = 0;
1133 
1134 	if (newp)
1135 		return (EPERM);
1136 
1137 	SYSCTL_SCALAR_CORE_TYP(oldp, oldlenp, &val, quad_t)
1138 
1139 	return (error);
1140 }
1141 
1142 /*
1143  * Validate parameters and get old / set new parameters
1144  * for a string-valued sysctl function.
1145  */
1146 int
1147 sysctl_string(void *oldp, size_t *oldlenp, void *newp, size_t newlen, char *str,
1148     size_t maxlen)
1149 {
1150 	int error = 0, err2 = 0;
1151 	size_t len;
1152 
1153 	if (newp && newlen >= maxlen)
1154 		return (EINVAL);
1155 
1156 	SYSCTL_STRING_CORE(oldp, oldlenp, str);
1157 
1158 	if (error == 0 && newp) {
1159 		error = copyin(newp, str, newlen);
1160 		str[newlen] = 0;
1161 	}
1162 	return (error);
1163 }
1164 
1165 /*
1166  * As above, but read-only.
1167  */
1168 int
1169 sysctl_rdstring(void *oldp, size_t *oldlenp, void *newp, const char *str)
1170 {
1171 	int error = 0, err2 = 0;
1172 	size_t len;
1173 
1174 	if (newp)
1175 		return (EPERM);
1176 
1177 	SYSCTL_STRING_CORE(oldp, oldlenp, str);
1178 
1179 	return (error);
1180 }
1181 
1182 /*
1183  * Validate parameters and get old / set new parameters
1184  * for a structure oriented sysctl function.
1185  */
1186 int
1187 sysctl_struct(void *oldp, size_t *oldlenp, void *newp, size_t newlen, void *sp,
1188     size_t len)
1189 {
1190 	int error = 0;
1191 
1192 	SYSCTL_SCALAR_NEWPCHECK_LEN(newp, newlen, len)
1193 	SYSCTL_SCALAR_CORE_LEN(oldp, oldlenp, sp, len)
1194 	SYSCTL_SCALAR_NEWPCOP_LEN(newp, sp, len)
1195 
1196 	return (error);
1197 }
1198 
1199 /*
1200  * Validate parameters and get old parameters
1201  * for a structure oriented sysctl function.
1202  */
1203 int
1204 sysctl_rdstruct(void *oldp, size_t *oldlenp, void *newp, const void *sp,
1205     size_t len)
1206 {
1207 	int error = 0;
1208 
1209 	if (newp)
1210 		return (EPERM);
1211 
1212 	SYSCTL_SCALAR_CORE_LEN(oldp, oldlenp, sp, len)
1213 
1214 	return (error);
1215 }
1216 
1217 /*
1218  * As above, but can return a truncated result.
1219  */
1220 int
1221 sysctl_rdminstruct(void *oldp, size_t *oldlenp, void *newp, const void *sp,
1222     size_t len)
1223 {
1224 	int error = 0;
1225 
1226 	if (newp)
1227 		return (EPERM);
1228 
1229 	len = min(*oldlenp, len);
1230 	SYSCTL_SCALAR_CORE_LEN(oldp, oldlenp, sp, len)
1231 
1232 	return (error);
1233 }
1234 
1235 /*
1236  * Get file structures.
1237  */
1238 static int
1239 sysctl_file(void *vwhere, size_t *sizep)
1240 {
1241 	int error;
1242 	size_t buflen;
1243 	struct file *fp;
1244 	char *start, *where;
1245 
1246 	start = where = vwhere;
1247 	buflen = *sizep;
1248 	if (where == NULL) {
1249 		/*
1250 		 * overestimate by 10 files
1251 		 */
1252 		*sizep = sizeof(filehead) + (nfiles + 10) * sizeof(struct file);
1253 		return (0);
1254 	}
1255 
1256 	/*
1257 	 * first copyout filehead
1258 	 */
1259 	if (buflen < sizeof(filehead)) {
1260 		*sizep = 0;
1261 		return (0);
1262 	}
1263 	error = copyout((caddr_t)&filehead, where, sizeof(filehead));
1264 	if (error)
1265 		return (error);
1266 	buflen -= sizeof(filehead);
1267 	where += sizeof(filehead);
1268 
1269 	/*
1270 	 * followed by an array of file structures
1271 	 */
1272 	LIST_FOREACH(fp, &filehead, f_list) {
1273 		if (buflen < sizeof(struct file)) {
1274 			*sizep = where - start;
1275 			return (ENOMEM);
1276 		}
1277 		error = copyout((caddr_t)fp, where, sizeof(struct file));
1278 		if (error)
1279 			return (error);
1280 		buflen -= sizeof(struct file);
1281 		where += sizeof(struct file);
1282 	}
1283 	*sizep = where - start;
1284 	return (0);
1285 }
1286 
1287 #if defined(SYSVMSG) || defined(SYSVSEM) || defined(SYSVSHM)
1288 #define	FILL_PERM(src, dst) do { \
1289 	(dst)._key = (src)._key; \
1290 	(dst).uid = (src).uid; \
1291 	(dst).gid = (src).gid; \
1292 	(dst).cuid = (src).cuid; \
1293 	(dst).cgid = (src).cgid; \
1294 	(dst).mode = (src).mode; \
1295 	(dst)._seq = (src)._seq; \
1296 } while (/*CONSTCOND*/ 0);
1297 #define	FILL_MSG(src, dst) do { \
1298 	FILL_PERM((src).msg_perm, (dst).msg_perm); \
1299 	(dst).msg_qnum = (src).msg_qnum; \
1300 	(dst).msg_qbytes = (src).msg_qbytes; \
1301 	(dst)._msg_cbytes = (src)._msg_cbytes; \
1302 	(dst).msg_lspid = (src).msg_lspid; \
1303 	(dst).msg_lrpid = (src).msg_lrpid; \
1304 	(dst).msg_stime = (src).msg_stime; \
1305 	(dst).msg_rtime = (src).msg_rtime; \
1306 	(dst).msg_ctime = (src).msg_ctime; \
1307 } while (/*CONSTCOND*/ 0)
1308 #define	FILL_SEM(src, dst) do { \
1309 	FILL_PERM((src).sem_perm, (dst).sem_perm); \
1310 	(dst).sem_nsems = (src).sem_nsems; \
1311 	(dst).sem_otime = (src).sem_otime; \
1312 	(dst).sem_ctime = (src).sem_ctime; \
1313 } while (/*CONSTCOND*/ 0)
1314 #define	FILL_SHM(src, dst) do { \
1315 	FILL_PERM((src).shm_perm, (dst).shm_perm); \
1316 	(dst).shm_segsz = (src).shm_segsz; \
1317 	(dst).shm_lpid = (src).shm_lpid; \
1318 	(dst).shm_cpid = (src).shm_cpid; \
1319 	(dst).shm_atime = (src).shm_atime; \
1320 	(dst).shm_dtime = (src).shm_dtime; \
1321 	(dst).shm_ctime = (src).shm_ctime; \
1322 	(dst).shm_nattch = (src).shm_nattch; \
1323 } while (/*CONSTCOND*/ 0)
1324 
1325 static int
1326 sysctl_sysvipc(int *name, u_int namelen, void *where, size_t *sizep)
1327 {
1328 #ifdef SYSVMSG
1329 	struct msg_sysctl_info *msgsi = NULL;
1330 #endif
1331 #ifdef SYSVSEM
1332 	struct sem_sysctl_info *semsi = NULL;
1333 #endif
1334 #ifdef SYSVSHM
1335 	struct shm_sysctl_info *shmsi = NULL;
1336 #endif
1337 	size_t infosize, dssize, tsize, buflen;
1338 	void *buf = NULL;
1339 	char *start;
1340 	int32_t nds;
1341 	int i, error, ret;
1342 
1343 	if (namelen != 1)
1344 		return (EINVAL);
1345 
1346 	start = where;
1347 	buflen = *sizep;
1348 
1349 	switch (*name) {
1350 	case KERN_SYSVIPC_MSG_INFO:
1351 #ifdef SYSVMSG
1352 		infosize = sizeof(msgsi->msginfo);
1353 		nds = msginfo.msgmni;
1354 		dssize = sizeof(msgsi->msgids[0]);
1355 		break;
1356 #else
1357 		return (EINVAL);
1358 #endif
1359 	case KERN_SYSVIPC_SEM_INFO:
1360 #ifdef SYSVSEM
1361 		infosize = sizeof(semsi->seminfo);
1362 		nds = seminfo.semmni;
1363 		dssize = sizeof(semsi->semids[0]);
1364 		break;
1365 #else
1366 		return (EINVAL);
1367 #endif
1368 	case KERN_SYSVIPC_SHM_INFO:
1369 #ifdef SYSVSHM
1370 		infosize = sizeof(shmsi->shminfo);
1371 		nds = shminfo.shmmni;
1372 		dssize = sizeof(shmsi->shmids[0]);
1373 		break;
1374 #else
1375 		return (EINVAL);
1376 #endif
1377 	default:
1378 		return (EINVAL);
1379 	}
1380 	/*
1381 	 * Round infosize to 64 bit boundary if requesting more than just
1382 	 * the info structure or getting the total data size.
1383 	 */
1384 	if (where == NULL || *sizep > infosize)
1385 		infosize = ((infosize + 7) / 8) * 8;
1386 	tsize = infosize + nds * dssize;
1387 
1388 	/* Return just the total size required. */
1389 	if (where == NULL) {
1390 		*sizep = tsize;
1391 		return (0);
1392 	}
1393 
1394 	/* Not enough room for even the info struct. */
1395 	if (buflen < infosize) {
1396 		*sizep = 0;
1397 		return (ENOMEM);
1398 	}
1399 	buf = malloc(min(tsize, buflen), M_TEMP, M_WAITOK);
1400 	memset(buf, 0, min(tsize, buflen));
1401 
1402 	switch (*name) {
1403 #ifdef SYSVMSG
1404 	case KERN_SYSVIPC_MSG_INFO:
1405 		msgsi = (struct msg_sysctl_info *)buf;
1406 		msgsi->msginfo = msginfo;
1407 		break;
1408 #endif
1409 #ifdef SYSVSEM
1410 	case KERN_SYSVIPC_SEM_INFO:
1411 		semsi = (struct sem_sysctl_info *)buf;
1412 		semsi->seminfo = seminfo;
1413 		break;
1414 #endif
1415 #ifdef SYSVSHM
1416 	case KERN_SYSVIPC_SHM_INFO:
1417 		shmsi = (struct shm_sysctl_info *)buf;
1418 		shmsi->shminfo = shminfo;
1419 		break;
1420 #endif
1421 	}
1422 	buflen -= infosize;
1423 
1424 	ret = 0;
1425 	if (buflen > 0) {
1426 		/* Fill in the IPC data structures.  */
1427 		for (i = 0; i < nds; i++) {
1428 			if (buflen < dssize) {
1429 				ret = ENOMEM;
1430 				break;
1431 			}
1432 			switch (*name) {
1433 #ifdef SYSVMSG
1434 			case KERN_SYSVIPC_MSG_INFO:
1435 				FILL_MSG(msqids[i], msgsi->msgids[i]);
1436 				break;
1437 #endif
1438 #ifdef SYSVSEM
1439 			case KERN_SYSVIPC_SEM_INFO:
1440 				FILL_SEM(sema[i], semsi->semids[i]);
1441 				break;
1442 #endif
1443 #ifdef SYSVSHM
1444 			case KERN_SYSVIPC_SHM_INFO:
1445 				FILL_SHM(shmsegs[i], shmsi->shmids[i]);
1446 				break;
1447 #endif
1448 			}
1449 			buflen -= dssize;
1450 		}
1451 	}
1452 	*sizep -= buflen;
1453 	error = copyout(buf, start, *sizep);
1454 	/* If copyout succeeded, use return code set earlier. */
1455 	if (error == 0)
1456 		error = ret;
1457 	if (buf)
1458 		free(buf, M_TEMP);
1459 	return (error);
1460 }
1461 #endif /* SYSVMSG || SYSVSEM || SYSVSHM */
1462 
1463 static int
1464 sysctl_msgbuf(void *vwhere, size_t *sizep)
1465 {
1466 	char *where = vwhere;
1467 	size_t len, maxlen = *sizep;
1468 	long beg, end;
1469 	int error;
1470 
1471 	/*
1472 	 * deal with cases where the message buffer has
1473 	 * become corrupted.
1474 	 */
1475 	if (!msgbufenabled || msgbufp->msg_magic != MSG_MAGIC) {
1476 		msgbufenabled = 0;
1477 		return (ENXIO);
1478 	}
1479 
1480 	if (where == NULL) {
1481 		/* always return full buffer size */
1482 		*sizep = msgbufp->msg_bufs;
1483 		return (0);
1484 	}
1485 
1486 	error = 0;
1487 	maxlen = min(msgbufp->msg_bufs, maxlen);
1488 
1489 	/*
1490 	 * First, copy from the write pointer to the end of
1491 	 * message buffer.
1492 	 */
1493 	beg = msgbufp->msg_bufx;
1494 	end = msgbufp->msg_bufs;
1495 	while (maxlen > 0) {
1496 		len = min(end - beg, maxlen);
1497 		if (len == 0)
1498 			break;
1499 		error = copyout(&msgbufp->msg_bufc[beg], where, len);
1500 		if (error)
1501 			break;
1502 		where += len;
1503 		maxlen -= len;
1504 
1505 		/*
1506 		 * ... then, copy from the beginning of message buffer to
1507 		 * the write pointer.
1508 		 */
1509 		beg = 0;
1510 		end = msgbufp->msg_bufx;
1511 	}
1512 	return (error);
1513 }
1514 
1515 /*
1516  * try over estimating by 5 procs
1517  */
1518 #define KERN_PROCSLOP	(5 * sizeof(struct kinfo_proc))
1519 
1520 static int
1521 sysctl_doeproc(int *name, u_int namelen, void *vwhere, size_t *sizep)
1522 {
1523 	struct eproc eproc;
1524 	struct kinfo_proc2 kproc2;
1525 	struct kinfo_proc *dp;
1526 	struct proc *p;
1527 	const struct proclist_desc *pd;
1528 	char *where, *dp2;
1529 	int type, op, arg;
1530 	u_int elem_size, elem_count;
1531 	size_t buflen, needed;
1532 	int error;
1533 
1534 	dp = vwhere;
1535 	dp2 = where = vwhere;
1536 	buflen = where != NULL ? *sizep : 0;
1537 	error = 0;
1538 	needed = 0;
1539 	type = name[0];
1540 
1541 	if (type == KERN_PROC) {
1542 		if (namelen != 3 && !(namelen == 2 && name[1] == KERN_PROC_ALL))
1543 			return (EINVAL);
1544 		op = name[1];
1545 		if (op != KERN_PROC_ALL)
1546 			arg = name[2];
1547 		else
1548 			arg = 0;		/* Quell compiler warning */
1549 		elem_size = elem_count = 0;	/* Ditto */
1550 	} else {
1551 		if (namelen != 5)
1552 			return (EINVAL);
1553 		op = name[1];
1554 		arg = name[2];
1555 		elem_size = name[3];
1556 		elem_count = name[4];
1557 	}
1558 
1559 	proclist_lock_read();
1560 
1561 	pd = proclists;
1562 again:
1563 	for (p = LIST_FIRST(pd->pd_list); p != NULL; p = LIST_NEXT(p, p_list)) {
1564 		/*
1565 		 * Skip embryonic processes.
1566 		 */
1567 		if (p->p_stat == SIDL)
1568 			continue;
1569 		/*
1570 		 * TODO - make more efficient (see notes below).
1571 		 * do by session.
1572 		 */
1573 		switch (op) {
1574 
1575 		case KERN_PROC_PID:
1576 			/* could do this with just a lookup */
1577 			if (p->p_pid != (pid_t)arg)
1578 				continue;
1579 			break;
1580 
1581 		case KERN_PROC_PGRP:
1582 			/* could do this by traversing pgrp */
1583 			if (p->p_pgrp->pg_id != (pid_t)arg)
1584 				continue;
1585 			break;
1586 
1587 		case KERN_PROC_SESSION:
1588 			if (p->p_session->s_sid != (pid_t)arg)
1589 				continue;
1590 			break;
1591 
1592 		case KERN_PROC_TTY:
1593 			if (arg == (int) KERN_PROC_TTY_REVOKE) {
1594 				if ((p->p_flag & P_CONTROLT) == 0 ||
1595 				    p->p_session->s_ttyp == NULL ||
1596 				    p->p_session->s_ttyvp != NULL)
1597 					continue;
1598 			} else if ((p->p_flag & P_CONTROLT) == 0 ||
1599 			    p->p_session->s_ttyp == NULL) {
1600 				if ((dev_t)arg != KERN_PROC_TTY_NODEV)
1601 					continue;
1602 			} else if (p->p_session->s_ttyp->t_dev != (dev_t)arg)
1603 				continue;
1604 			break;
1605 
1606 		case KERN_PROC_UID:
1607 			if (p->p_ucred->cr_uid != (uid_t)arg)
1608 				continue;
1609 			break;
1610 
1611 		case KERN_PROC_RUID:
1612 			if (p->p_cred->p_ruid != (uid_t)arg)
1613 				continue;
1614 			break;
1615 
1616 		case KERN_PROC_GID:
1617 			if (p->p_ucred->cr_gid != (uid_t)arg)
1618 				continue;
1619 			break;
1620 
1621 		case KERN_PROC_RGID:
1622 			if (p->p_cred->p_rgid != (uid_t)arg)
1623 				continue;
1624 			break;
1625 
1626 		case KERN_PROC_ALL:
1627 			/* allow everything */
1628 			break;
1629 
1630 		default:
1631 			error = EINVAL;
1632 			goto cleanup;
1633 		}
1634 		if (type == KERN_PROC) {
1635 			if (buflen >= sizeof(struct kinfo_proc)) {
1636 				fill_eproc(p, &eproc);
1637 				error = copyout((caddr_t)p, &dp->kp_proc,
1638 				    sizeof(struct proc));
1639 				if (error)
1640 					goto cleanup;
1641 				error = copyout((caddr_t)&eproc, &dp->kp_eproc,
1642 				    sizeof(eproc));
1643 				if (error)
1644 					goto cleanup;
1645 				dp++;
1646 				buflen -= sizeof(struct kinfo_proc);
1647 			}
1648 			needed += sizeof(struct kinfo_proc);
1649 		} else { /* KERN_PROC2 */
1650 			if (buflen >= elem_size && elem_count > 0) {
1651 				fill_kproc2(p, &kproc2);
1652 				/*
1653 				 * Copy out elem_size, but not larger than
1654 				 * the size of a struct kinfo_proc2.
1655 				 */
1656 				error = copyout(&kproc2, dp2,
1657 				    min(sizeof(kproc2), elem_size));
1658 				if (error)
1659 					goto cleanup;
1660 				dp2 += elem_size;
1661 				buflen -= elem_size;
1662 				elem_count--;
1663 			}
1664 			needed += elem_size;
1665 		}
1666 	}
1667 	pd++;
1668 	if (pd->pd_list != NULL)
1669 		goto again;
1670 	proclist_unlock_read();
1671 
1672 	if (where != NULL) {
1673 		if (type == KERN_PROC)
1674 			*sizep = (caddr_t)dp - where;
1675 		else
1676 			*sizep = dp2 - where;
1677 		if (needed > *sizep)
1678 			return (ENOMEM);
1679 	} else {
1680 		needed += KERN_PROCSLOP;
1681 		*sizep = needed;
1682 	}
1683 	return (0);
1684  cleanup:
1685 	proclist_unlock_read();
1686 	return (error);
1687 }
1688 
1689 
1690 /*
1691  * try over estimating by 5 LWPs
1692  */
1693 #define KERN_LWPSLOP	(5 * sizeof(struct kinfo_lwp))
1694 
1695 static int
1696 sysctl_dolwp(int *name, u_int namelen, void *vwhere, size_t *sizep)
1697 {
1698 	struct kinfo_lwp klwp;
1699 	struct proc *p;
1700 	struct lwp *l;
1701 	char *where, *dp;
1702 	int type, pid, elem_size, elem_count;
1703 	int buflen, needed, error;
1704 
1705 	dp = where = vwhere;
1706 	buflen = where != NULL ? *sizep : 0;
1707 	error = needed = 0;
1708 	type = name[0];
1709 
1710 	if (namelen != 4)
1711 		return (EINVAL);
1712 	pid = name[1];
1713 	elem_size = name[2];
1714 	elem_count = name[3];
1715 
1716 	p = pfind(pid);
1717 	if (p == NULL)
1718 		return (ESRCH);
1719 	LIST_FOREACH(l, &p->p_lwps, l_sibling) {
1720 		if (buflen >= elem_size && elem_count > 0) {
1721 			fill_lwp(l, &klwp);
1722 			/*
1723 			 * Copy out elem_size, but not larger than
1724 			 * the size of a struct kinfo_proc2.
1725 			 */
1726 			error = copyout(&klwp, dp,
1727 			    min(sizeof(klwp), elem_size));
1728 			if (error)
1729 				goto cleanup;
1730 			dp += elem_size;
1731 			buflen -= elem_size;
1732 			elem_count--;
1733 		}
1734 		needed += elem_size;
1735 	}
1736 
1737 	if (where != NULL) {
1738 		*sizep = dp - where;
1739 		if (needed > *sizep)
1740 			return (ENOMEM);
1741 	} else {
1742 		needed += KERN_PROCSLOP;
1743 		*sizep = needed;
1744 	}
1745 	return (0);
1746  cleanup:
1747 	return (error);
1748 }
1749 
1750 /*
1751  * Fill in an eproc structure for the specified process.
1752  */
1753 void
1754 fill_eproc(struct proc *p, struct eproc *ep)
1755 {
1756 	struct tty *tp;
1757 	struct lwp *l;
1758 
1759 	ep->e_paddr = p;
1760 	ep->e_sess = p->p_session;
1761 	ep->e_pcred = *p->p_cred;
1762 	ep->e_ucred = *p->p_ucred;
1763 	if (p->p_stat == SIDL || P_ZOMBIE(p)) {
1764 		ep->e_vm.vm_rssize = 0;
1765 		ep->e_vm.vm_tsize = 0;
1766 		ep->e_vm.vm_dsize = 0;
1767 		ep->e_vm.vm_ssize = 0;
1768 		/* ep->e_vm.vm_pmap = XXX; */
1769 	} else {
1770 		struct vmspace *vm = p->p_vmspace;
1771 
1772 		ep->e_vm.vm_rssize = vm_resident_count(vm);
1773 		ep->e_vm.vm_tsize = vm->vm_tsize;
1774 		ep->e_vm.vm_dsize = vm->vm_dsize;
1775 		ep->e_vm.vm_ssize = vm->vm_ssize;
1776 
1777 		/* Pick a "representative" LWP */
1778 		l = proc_representative_lwp(p);
1779 
1780 		if (l->l_wmesg)
1781 			strncpy(ep->e_wmesg, l->l_wmesg, WMESGLEN);
1782 	}
1783 	if (p->p_pptr)
1784 		ep->e_ppid = p->p_pptr->p_pid;
1785 	else
1786 		ep->e_ppid = 0;
1787 	ep->e_pgid = p->p_pgrp->pg_id;
1788 	ep->e_sid = ep->e_sess->s_sid;
1789 	ep->e_jobc = p->p_pgrp->pg_jobc;
1790 	if ((p->p_flag & P_CONTROLT) &&
1791 	    (tp = ep->e_sess->s_ttyp)) {
1792 		ep->e_tdev = tp->t_dev;
1793 		ep->e_tpgid = tp->t_pgrp ? tp->t_pgrp->pg_id : NO_PGID;
1794 		ep->e_tsess = tp->t_session;
1795 	} else
1796 		ep->e_tdev = NODEV;
1797 
1798 	ep->e_xsize = ep->e_xrssize = 0;
1799 	ep->e_xccount = ep->e_xswrss = 0;
1800 	ep->e_flag = ep->e_sess->s_ttyvp ? EPROC_CTTY : 0;
1801 	if (SESS_LEADER(p))
1802 		ep->e_flag |= EPROC_SLEADER;
1803 	strncpy(ep->e_login, ep->e_sess->s_login, MAXLOGNAME);
1804 }
1805 
1806 /*
1807  * Fill in an eproc structure for the specified process.
1808  */
1809 static void
1810 fill_kproc2(struct proc *p, struct kinfo_proc2 *ki)
1811 {
1812 	struct tty *tp;
1813 	struct lwp *l;
1814 	struct timeval ut, st;
1815 
1816 	memset(ki, 0, sizeof(*ki));
1817 
1818 	ki->p_paddr = PTRTOINT64(p);
1819 	ki->p_fd = PTRTOINT64(p->p_fd);
1820 	ki->p_cwdi = PTRTOINT64(p->p_cwdi);
1821 	ki->p_stats = PTRTOINT64(p->p_stats);
1822 	ki->p_limit = PTRTOINT64(p->p_limit);
1823 	ki->p_vmspace = PTRTOINT64(p->p_vmspace);
1824 	ki->p_sigacts = PTRTOINT64(p->p_sigacts);
1825 	ki->p_sess = PTRTOINT64(p->p_session);
1826 	ki->p_tsess = 0;	/* may be changed if controlling tty below */
1827 	ki->p_ru = PTRTOINT64(p->p_ru);
1828 
1829 	ki->p_eflag = 0;
1830 	ki->p_exitsig = p->p_exitsig;
1831 	ki->p_flag = p->p_flag;
1832 
1833 	ki->p_pid = p->p_pid;
1834 	if (p->p_pptr)
1835 		ki->p_ppid = p->p_pptr->p_pid;
1836 	else
1837 		ki->p_ppid = 0;
1838 	ki->p_sid = p->p_session->s_sid;
1839 	ki->p__pgid = p->p_pgrp->pg_id;
1840 
1841 	ki->p_tpgid = NO_PGID;	/* may be changed if controlling tty below */
1842 
1843 	ki->p_uid = p->p_ucred->cr_uid;
1844 	ki->p_ruid = p->p_cred->p_ruid;
1845 	ki->p_gid = p->p_ucred->cr_gid;
1846 	ki->p_rgid = p->p_cred->p_rgid;
1847 	ki->p_svuid = p->p_cred->p_svuid;
1848 	ki->p_svgid = p->p_cred->p_svgid;
1849 
1850 	memcpy(ki->p_groups, p->p_cred->pc_ucred->cr_groups,
1851 	    min(sizeof(ki->p_groups), sizeof(p->p_cred->pc_ucred->cr_groups)));
1852 	ki->p_ngroups = p->p_cred->pc_ucred->cr_ngroups;
1853 
1854 	ki->p_jobc = p->p_pgrp->pg_jobc;
1855 	if ((p->p_flag & P_CONTROLT) && (tp = p->p_session->s_ttyp)) {
1856 		ki->p_tdev = tp->t_dev;
1857 		ki->p_tpgid = tp->t_pgrp ? tp->t_pgrp->pg_id : NO_PGID;
1858 		ki->p_tsess = PTRTOINT64(tp->t_session);
1859 	} else {
1860 		ki->p_tdev = NODEV;
1861 	}
1862 
1863 	ki->p_estcpu = p->p_estcpu;
1864 	ki->p_rtime_sec = p->p_rtime.tv_sec;
1865 	ki->p_rtime_usec = p->p_rtime.tv_usec;
1866 	ki->p_cpticks = p->p_cpticks;
1867 	ki->p_pctcpu = p->p_pctcpu;
1868 
1869 	ki->p_uticks = p->p_uticks;
1870 	ki->p_sticks = p->p_sticks;
1871 	ki->p_iticks = p->p_iticks;
1872 
1873 	ki->p_tracep = PTRTOINT64(p->p_tracep);
1874 	ki->p_traceflag = p->p_traceflag;
1875 
1876 
1877 	memcpy(&ki->p_siglist, &p->p_sigctx.ps_siglist, sizeof(ki_sigset_t));
1878 	memcpy(&ki->p_sigmask, &p->p_sigctx.ps_sigmask, sizeof(ki_sigset_t));
1879 	memcpy(&ki->p_sigignore, &p->p_sigctx.ps_sigignore,sizeof(ki_sigset_t));
1880 	memcpy(&ki->p_sigcatch, &p->p_sigctx.ps_sigcatch, sizeof(ki_sigset_t));
1881 
1882 	ki->p_stat = p->p_stat; /* Will likely be overridden by LWP status */
1883 	ki->p_realstat = p->p_stat;
1884 	ki->p_nice = p->p_nice;
1885 
1886 	ki->p_xstat = p->p_xstat;
1887 	ki->p_acflag = p->p_acflag;
1888 
1889 	strncpy(ki->p_comm, p->p_comm,
1890 	    min(sizeof(ki->p_comm), sizeof(p->p_comm)));
1891 
1892 	strncpy(ki->p_login, p->p_session->s_login,
1893 	    min(sizeof ki->p_login - 1, sizeof p->p_session->s_login));
1894 
1895 	ki->p_nlwps = p->p_nlwps;
1896 	ki->p_nrlwps = p->p_nrlwps;
1897 	ki->p_realflag = p->p_flag;
1898 
1899 	if (p->p_stat == SIDL || P_ZOMBIE(p)) {
1900 		ki->p_vm_rssize = 0;
1901 		ki->p_vm_tsize = 0;
1902 		ki->p_vm_dsize = 0;
1903 		ki->p_vm_ssize = 0;
1904 		l = NULL;
1905 	} else {
1906 		struct vmspace *vm = p->p_vmspace;
1907 
1908 		ki->p_vm_rssize = vm_resident_count(vm);
1909 		ki->p_vm_tsize = vm->vm_tsize;
1910 		ki->p_vm_dsize = vm->vm_dsize;
1911 		ki->p_vm_ssize = vm->vm_ssize;
1912 
1913 		/* Pick a "representative" LWP */
1914 		l = proc_representative_lwp(p);
1915 		ki->p_forw = PTRTOINT64(l->l_forw);
1916 		ki->p_back = PTRTOINT64(l->l_back);
1917 		ki->p_addr = PTRTOINT64(l->l_addr);
1918 		ki->p_stat = l->l_stat;
1919 		ki->p_flag |= l->l_flag;
1920 		ki->p_swtime = l->l_swtime;
1921 		ki->p_slptime = l->l_slptime;
1922 		if (l->l_stat == LSONPROC) {
1923 			KDASSERT(l->l_cpu != NULL);
1924 			ki->p_schedflags = l->l_cpu->ci_schedstate.spc_flags;
1925 		} else
1926 			ki->p_schedflags = 0;
1927 		ki->p_holdcnt = l->l_holdcnt;
1928 		ki->p_priority = l->l_priority;
1929 		ki->p_usrpri = l->l_usrpri;
1930 		if (l->l_wmesg)
1931 			strncpy(ki->p_wmesg, l->l_wmesg, sizeof(ki->p_wmesg));
1932 		ki->p_wchan = PTRTOINT64(l->l_wchan);
1933 
1934 	}
1935 
1936 	if (p->p_session->s_ttyvp)
1937 		ki->p_eflag |= EPROC_CTTY;
1938 	if (SESS_LEADER(p))
1939 		ki->p_eflag |= EPROC_SLEADER;
1940 
1941 	/* XXX Is this double check necessary? */
1942 	if (P_ZOMBIE(p)) {
1943 		ki->p_uvalid = 0;
1944 	} else {
1945 		ki->p_uvalid = 1;
1946 
1947 		ki->p_ustart_sec = p->p_stats->p_start.tv_sec;
1948 		ki->p_ustart_usec = p->p_stats->p_start.tv_usec;
1949 
1950 		calcru(p, &ut, &st, 0);
1951 		ki->p_uutime_sec = ut.tv_sec;
1952 		ki->p_uutime_usec = ut.tv_usec;
1953 		ki->p_ustime_sec = st.tv_sec;
1954 		ki->p_ustime_usec = st.tv_usec;
1955 
1956 		ki->p_uru_maxrss = p->p_stats->p_ru.ru_maxrss;
1957 		ki->p_uru_ixrss = p->p_stats->p_ru.ru_ixrss;
1958 		ki->p_uru_idrss = p->p_stats->p_ru.ru_idrss;
1959 		ki->p_uru_isrss = p->p_stats->p_ru.ru_isrss;
1960 		ki->p_uru_minflt = p->p_stats->p_ru.ru_minflt;
1961 		ki->p_uru_majflt = p->p_stats->p_ru.ru_majflt;
1962 		ki->p_uru_nswap = p->p_stats->p_ru.ru_nswap;
1963 		ki->p_uru_inblock = p->p_stats->p_ru.ru_inblock;
1964 		ki->p_uru_oublock = p->p_stats->p_ru.ru_oublock;
1965 		ki->p_uru_msgsnd = p->p_stats->p_ru.ru_msgsnd;
1966 		ki->p_uru_msgrcv = p->p_stats->p_ru.ru_msgrcv;
1967 		ki->p_uru_nsignals = p->p_stats->p_ru.ru_nsignals;
1968 		ki->p_uru_nvcsw = p->p_stats->p_ru.ru_nvcsw;
1969 		ki->p_uru_nivcsw = p->p_stats->p_ru.ru_nivcsw;
1970 
1971 		timeradd(&p->p_stats->p_cru.ru_utime,
1972 			 &p->p_stats->p_cru.ru_stime, &ut);
1973 		ki->p_uctime_sec = ut.tv_sec;
1974 		ki->p_uctime_usec = ut.tv_usec;
1975 	}
1976 #ifdef MULTIPROCESSOR
1977 	if (l && l->l_cpu != NULL)
1978 		ki->p_cpuid = l->l_cpu->ci_cpuid;
1979 	else
1980 #endif
1981 		ki->p_cpuid = KI_NOCPU;
1982 
1983 }
1984 
1985 /*
1986  * Fill in a kinfo_lwp structure for the specified lwp.
1987  */
1988 static void
1989 fill_lwp(struct lwp *l, struct kinfo_lwp *kl)
1990 {
1991 
1992 	kl->l_forw = PTRTOINT64(l->l_forw);
1993 	kl->l_back = PTRTOINT64(l->l_back);
1994 	kl->l_laddr = PTRTOINT64(l);
1995 	kl->l_addr = PTRTOINT64(l->l_addr);
1996 	kl->l_stat = l->l_stat;
1997 	kl->l_lid = l->l_lid;
1998 	kl->l_flag = l->l_flag;
1999 
2000 	kl->l_swtime = l->l_swtime;
2001 	kl->l_slptime = l->l_slptime;
2002 	if (l->l_stat == LSONPROC) {
2003 		KDASSERT(l->l_cpu != NULL);
2004 		kl->l_schedflags = l->l_cpu->ci_schedstate.spc_flags;
2005 	} else
2006 		kl->l_schedflags = 0;
2007 	kl->l_holdcnt = l->l_holdcnt;
2008 	kl->l_priority = l->l_priority;
2009 	kl->l_usrpri = l->l_usrpri;
2010 	if (l->l_wmesg)
2011 		strncpy(kl->l_wmesg, l->l_wmesg, sizeof(kl->l_wmesg));
2012 	kl->l_wchan = PTRTOINT64(l->l_wchan);
2013 #ifdef MULTIPROCESSOR
2014 	if (l->l_cpu != NULL)
2015 		kl->l_cpuid = l->l_cpu->ci_cpuid;
2016 	else
2017 #endif
2018 		kl->l_cpuid = KI_NOCPU;
2019 }
2020 
2021 int
2022 sysctl_procargs(int *name, u_int namelen, void *where, size_t *sizep,
2023     struct proc *up)
2024 {
2025 	struct ps_strings pss;
2026 	struct proc *p;
2027 	size_t len, upper_bound, xlen, i;
2028 	struct uio auio;
2029 	struct iovec aiov;
2030 	vaddr_t argv;
2031 	pid_t pid;
2032 	int nargv, type, error;
2033 	char *arg;
2034 	char *tmp;
2035 
2036 	if (namelen != 2)
2037 		return (EINVAL);
2038 	pid = name[0];
2039 	type = name[1];
2040 
2041 	switch (type) {
2042 	case KERN_PROC_ARGV:
2043 	case KERN_PROC_NARGV:
2044 	case KERN_PROC_ENV:
2045 	case KERN_PROC_NENV:
2046 		/* ok */
2047 		break;
2048 	default:
2049 		return (EINVAL);
2050 	}
2051 
2052 	/* check pid */
2053 	if ((p = pfind(pid)) == NULL)
2054 		return (EINVAL);
2055 
2056 	/* only root or same user change look at the environment */
2057 	if (type == KERN_PROC_ENV || type == KERN_PROC_NENV) {
2058 		if (up->p_ucred->cr_uid != 0) {
2059 			if (up->p_cred->p_ruid != p->p_cred->p_ruid ||
2060 			    up->p_cred->p_ruid != p->p_cred->p_svuid)
2061 				return (EPERM);
2062 		}
2063 	}
2064 
2065 	if (sizep != NULL && where == NULL) {
2066 		if (type == KERN_PROC_NARGV || type == KERN_PROC_NENV)
2067 			*sizep = sizeof (int);
2068 		else
2069 			*sizep = ARG_MAX;	/* XXX XXX XXX */
2070 		return (0);
2071 	}
2072 	if (where == NULL || sizep == NULL)
2073 		return (EINVAL);
2074 
2075 	/*
2076 	 * Zombies don't have a stack, so we can't read their psstrings.
2077 	 * System processes also don't have a user stack.
2078 	 */
2079 	if (P_ZOMBIE(p) || (p->p_flag & P_SYSTEM) != 0)
2080 		return (EINVAL);
2081 
2082 	/*
2083 	 * Lock the process down in memory.
2084 	 */
2085 	/* XXXCDC: how should locking work here? */
2086 	if ((p->p_flag & P_WEXIT) || (p->p_vmspace->vm_refcnt < 1))
2087 		return (EFAULT);
2088 
2089 	p->p_vmspace->vm_refcnt++;	/* XXX */
2090 
2091 	/*
2092 	 * Allocate a temporary buffer to hold the arguments.
2093 	 */
2094 	arg = malloc(PAGE_SIZE, M_TEMP, M_WAITOK);
2095 
2096 	/*
2097 	 * Read in the ps_strings structure.
2098 	 */
2099 	aiov.iov_base = &pss;
2100 	aiov.iov_len = sizeof(pss);
2101 	auio.uio_iov = &aiov;
2102 	auio.uio_iovcnt = 1;
2103 	auio.uio_offset = (vaddr_t)p->p_psstr;
2104 	auio.uio_resid = sizeof(pss);
2105 	auio.uio_segflg = UIO_SYSSPACE;
2106 	auio.uio_rw = UIO_READ;
2107 	auio.uio_procp = NULL;
2108 	error = uvm_io(&p->p_vmspace->vm_map, &auio);
2109 	if (error)
2110 		goto done;
2111 
2112 	if (type == KERN_PROC_ARGV || type == KERN_PROC_NARGV)
2113 		memcpy(&nargv, (char *)&pss + p->p_psnargv, sizeof(nargv));
2114 	else
2115 		memcpy(&nargv, (char *)&pss + p->p_psnenv, sizeof(nargv));
2116 	if (type == KERN_PROC_NARGV || type == KERN_PROC_NENV) {
2117 		error = copyout(&nargv, where, sizeof(nargv));
2118 		*sizep = sizeof(nargv);
2119 		goto done;
2120 	}
2121 	/*
2122 	 * Now read the address of the argument vector.
2123 	 */
2124 	switch (type) {
2125 	case KERN_PROC_ARGV:
2126 		/* XXX compat32 stuff here */
2127 		memcpy(&tmp, (char *)&pss + p->p_psargv, sizeof(tmp));
2128 		break;
2129 	case KERN_PROC_ENV:
2130 		memcpy(&tmp, (char *)&pss + p->p_psenv, sizeof(tmp));
2131 		break;
2132 	default:
2133 		return (EINVAL);
2134 	}
2135 	auio.uio_offset = (off_t)(long)tmp;
2136 	aiov.iov_base = &argv;
2137 	aiov.iov_len = sizeof(argv);
2138 	auio.uio_iov = &aiov;
2139 	auio.uio_iovcnt = 1;
2140 	auio.uio_resid = sizeof(argv);
2141 	auio.uio_segflg = UIO_SYSSPACE;
2142 	auio.uio_rw = UIO_READ;
2143 	auio.uio_procp = NULL;
2144 	error = uvm_io(&p->p_vmspace->vm_map, &auio);
2145 	if (error)
2146 		goto done;
2147 
2148 	/*
2149 	 * Now copy in the actual argument vector, one page at a time,
2150 	 * since we don't know how long the vector is (though, we do
2151 	 * know how many NUL-terminated strings are in the vector).
2152 	 */
2153 	len = 0;
2154 	upper_bound = *sizep;
2155 	for (; nargv != 0 && len < upper_bound; len += xlen) {
2156 		aiov.iov_base = arg;
2157 		aiov.iov_len = PAGE_SIZE;
2158 		auio.uio_iov = &aiov;
2159 		auio.uio_iovcnt = 1;
2160 		auio.uio_offset = argv + len;
2161 		xlen = PAGE_SIZE - ((argv + len) & PAGE_MASK);
2162 		auio.uio_resid = xlen;
2163 		auio.uio_segflg = UIO_SYSSPACE;
2164 		auio.uio_rw = UIO_READ;
2165 		auio.uio_procp = NULL;
2166 		error = uvm_io(&p->p_vmspace->vm_map, &auio);
2167 		if (error)
2168 			goto done;
2169 
2170 		for (i = 0; i < xlen && nargv != 0; i++) {
2171 			if (arg[i] == '\0')
2172 				nargv--;	/* one full string */
2173 		}
2174 
2175 		/*
2176 		 * Make sure we don't copyout past the end of the user's
2177 		 * buffer.
2178 		 */
2179 		if (len + i > upper_bound)
2180 			i = upper_bound - len;
2181 
2182 		error = copyout(arg, (char *)where + len, i);
2183 		if (error)
2184 			break;
2185 
2186 		if (nargv == 0) {
2187 			len += i;
2188 			break;
2189 		}
2190 	}
2191 	*sizep = len;
2192 
2193 done:
2194 	uvmspace_free(p->p_vmspace);
2195 
2196 	free(arg, M_TEMP);
2197 	return (error);
2198 }
2199 
2200 #if NPTY > 0
2201 int pty_maxptys(int, int);		/* defined in kern/tty_pty.c */
2202 
2203 /*
2204  * Validate parameters and get old / set new parameters
2205  * for pty sysctl function.
2206  */
2207 static int
2208 sysctl_pty(void *oldp, size_t *oldlenp, void *newp, size_t newlen)
2209 {
2210 	int error = 0;
2211 	int oldmax = 0, newmax = 0;
2212 
2213 	/* get current value of maxptys */
2214 	oldmax = pty_maxptys(0, 0);
2215 
2216 	SYSCTL_SCALAR_CORE_TYP(oldp, oldlenp, &oldmax, int)
2217 
2218 	if (!error && newp) {
2219 		SYSCTL_SCALAR_NEWPCHECK_TYP(newp, newlen, int)
2220 		SYSCTL_SCALAR_NEWPCOP_TYP(newp, &newmax, int)
2221 
2222 		if (newmax != pty_maxptys(newmax, (newp != NULL)))
2223 			return (EINVAL);
2224 
2225 	}
2226 
2227 	return (error);
2228 }
2229 #endif /* NPTY > 0 */
2230 
2231 static int
2232 sysctl_dotkstat(int *name, u_int namelen, void *where, size_t *sizep,
2233     void *newp)
2234 {
2235 
2236 	/* all sysctl names at this level are terminal */
2237 	if (namelen != 1)
2238 		return (ENOTDIR);		/* overloaded */
2239 
2240 	switch (name[0]) {
2241 	case KERN_TKSTAT_NIN:
2242 		return (sysctl_rdquad(where, sizep, newp, tk_nin));
2243 	case KERN_TKSTAT_NOUT:
2244 		return (sysctl_rdquad(where, sizep, newp, tk_nout));
2245 	case KERN_TKSTAT_CANCC:
2246 		return (sysctl_rdquad(where, sizep, newp, tk_cancc));
2247 	case KERN_TKSTAT_RAWCC:
2248 		return (sysctl_rdquad(where, sizep, newp, tk_rawcc));
2249 	default:
2250 		return (EOPNOTSUPP);
2251 	}
2252 }
2253