xref: /netbsd-src/sys/kern/kern_sysctl.c (revision eb7c1594f145c931049e1fd9eb056a5987e87e59)
1 /*	$NetBSD: kern_sysctl.c,v 1.140 2003/08/07 16:31:50 agc 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. Neither the name of the University nor the names of its contributors
19  *    may be used to endorse or promote products derived from this software
20  *    without specific prior written permission.
21  *
22  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
23  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
24  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
25  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
26  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
27  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
28  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
29  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
30  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
31  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
32  * SUCH DAMAGE.
33  *
34  *	@(#)kern_sysctl.c	8.9 (Berkeley) 5/20/95
35  */
36 
37 /*
38  * sysctl system call.
39  */
40 
41 #include <sys/cdefs.h>
42 __KERNEL_RCSID(0, "$NetBSD: kern_sysctl.c,v 1.140 2003/08/07 16:31:50 agc Exp $");
43 
44 #include "opt_ddb.h"
45 #include "opt_insecure.h"
46 #include "opt_defcorename.h"
47 #include "opt_multiprocessor.h"
48 #include "opt_pipe.h"
49 #include "opt_sysv.h"
50 #include "pty.h"
51 #include "rnd.h"
52 
53 #include <sys/param.h>
54 #include <sys/systm.h>
55 #include <sys/kernel.h>
56 #include <sys/buf.h>
57 #include <sys/device.h>
58 #include <sys/disklabel.h>
59 #include <sys/dkstat.h>
60 #include <sys/exec.h>
61 #include <sys/file.h>
62 #include <sys/ioctl.h>
63 #include <sys/malloc.h>
64 #include <sys/mount.h>
65 #include <sys/msgbuf.h>
66 #include <sys/pool.h>
67 #include <sys/proc.h>
68 #include <sys/resource.h>
69 #include <sys/resourcevar.h>
70 #include <sys/sa.h>
71 #include <sys/syscallargs.h>
72 #include <sys/tty.h>
73 #include <sys/unistd.h>
74 #include <sys/vnode.h>
75 #include <sys/socketvar.h>
76 #define	__SYSCTL_PRIVATE
77 #include <sys/sysctl.h>
78 #include <sys/lock.h>
79 #include <sys/namei.h>
80 
81 #if defined(SYSVMSG) || defined(SYSVSEM) || defined(SYSVSHM)
82 #include <sys/ipc.h>
83 #endif
84 #ifdef SYSVMSG
85 #include <sys/msg.h>
86 #endif
87 #ifdef SYSVSEM
88 #include <sys/sem.h>
89 #endif
90 #ifdef SYSVSHM
91 #include <sys/shm.h>
92 #endif
93 
94 #include <dev/cons.h>
95 
96 #if defined(DDB)
97 #include <ddb/ddbvar.h>
98 #endif
99 
100 #ifndef PIPE_SOCKETPAIR
101 #include <sys/pipe.h>
102 #endif
103 
104 #if NRND > 0
105 #include <sys/rnd.h>
106 #endif
107 
108 #define PTRTOINT64(foo)	((u_int64_t)(uintptr_t)(foo))
109 
110 static int sysctl_file(void *, size_t *);
111 #if defined(SYSVMSG) || defined(SYSVSEM) || defined(SYSVSHM)
112 static int sysctl_sysvipc(int *, u_int, void *, size_t *);
113 #endif
114 static int sysctl_msgbuf(void *, size_t *);
115 static int sysctl_doeproc(int *, u_int, void *, size_t *);
116 static int sysctl_dolwp(int *, u_int, void *, size_t *);
117 static int sysctl_dotkstat(int *, u_int, void *, size_t *, void *);
118 #ifdef MULTIPROCESSOR
119 static int sysctl_docptime(void *, size_t *, void *);
120 static int sysctl_ncpus(void);
121 #endif
122 static void fill_kproc2(struct proc *, struct kinfo_proc2 *);
123 static void fill_lwp(struct lwp *, struct kinfo_lwp *);
124 static int sysctl_procargs(int *, u_int, void *, size_t *, struct proc *);
125 #if NPTY > 0
126 static int sysctl_pty(void *, size_t *, void *, size_t);
127 #endif
128 
129 /*
130  * The `sysctl_memlock' is intended to keep too many processes from
131  * locking down memory by doing sysctls at once.  Whether or not this
132  * is really a good idea to worry about it probably a subject of some
133  * debate.
134  */
135 struct lock sysctl_memlock;
136 
137 void
138 sysctl_init(void)
139 {
140 
141 	lockinit(&sysctl_memlock, PRIBIO|PCATCH, "sysctl", 0, 0);
142 }
143 
144 int
145 sys___sysctl(struct lwp *l, void *v, register_t *retval)
146 {
147 	struct sys___sysctl_args /* {
148 		syscallarg(int *) name;
149 		syscallarg(u_int) namelen;
150 		syscallarg(void *) old;
151 		syscallarg(size_t *) oldlenp;
152 		syscallarg(void *) new;
153 		syscallarg(size_t) newlen;
154 	} */ *uap = v;
155 	struct proc *p = l->l_proc;
156 	int error;
157 	size_t savelen = 0, oldlen = 0;
158 	sysctlfn *fn;
159 	int name[CTL_MAXNAME];
160 	size_t *oldlenp;
161 
162 	/*
163 	 * all top-level sysctl names are non-terminal
164 	 */
165 	if (SCARG(uap, namelen) > CTL_MAXNAME || SCARG(uap, namelen) < 2)
166 		return (EINVAL);
167 	error = copyin(SCARG(uap, name), &name,
168 	    SCARG(uap, namelen) * sizeof(int));
169 	if (error)
170 		return (error);
171 
172 	/*
173 	 * For all but CTL_PROC, must be root to change a value.
174 	 * For CTL_PROC, must be root, or owner of the proc (and not suid),
175 	 * this is checked in proc_sysctl() (once we know the targer proc).
176 	 */
177 	if (SCARG(uap, new) != NULL && name[0] != CTL_PROC &&
178 	    (error = suser(p->p_ucred, &p->p_acflag)))
179 		return (error);
180 
181 	switch (name[0]) {
182 	case CTL_KERN:
183 		fn = kern_sysctl;
184 		break;
185 	case CTL_HW:
186 		fn = hw_sysctl;
187 		break;
188 	case CTL_VM:
189 		fn = uvm_sysctl;
190 		break;
191 	case CTL_NET:
192 		fn = net_sysctl;
193 		break;
194 	case CTL_VFS:
195 		fn = vfs_sysctl;
196 		break;
197 	case CTL_MACHDEP:
198 		fn = cpu_sysctl;
199 		break;
200 #ifdef DEBUG
201 	case CTL_DEBUG:
202 		fn = debug_sysctl;
203 		break;
204 #endif
205 #ifdef DDB
206 	case CTL_DDB:
207 		fn = ddb_sysctl;
208 		break;
209 #endif
210 	case CTL_PROC:
211 		fn = proc_sysctl;
212 		break;
213 
214 	case CTL_EMUL:
215 		fn = emul_sysctl;
216 		break;
217 	default:
218 		return (EOPNOTSUPP);
219 	}
220 
221 	/*
222 	 * XXX Hey, we wire `old', but what about `new'?
223 	 */
224 
225 	oldlenp = SCARG(uap, oldlenp);
226 	if (oldlenp) {
227 		if ((error = copyin(oldlenp, &oldlen, sizeof(oldlen))))
228 			return (error);
229 		oldlenp = &oldlen;
230 	}
231 	if (SCARG(uap, old) != NULL) {
232 		error = lockmgr(&sysctl_memlock, LK_EXCLUSIVE, NULL);
233 		if (error)
234 			return (error);
235 		error = uvm_vslock(p, SCARG(uap, old), oldlen, VM_PROT_WRITE);
236 		if (error) {
237 			(void) lockmgr(&sysctl_memlock, LK_RELEASE, NULL);
238 			return (error);
239 		}
240 		savelen = oldlen;
241 	}
242 	error = (*fn)(name + 1, SCARG(uap, namelen) - 1, SCARG(uap, old),
243 	    oldlenp, SCARG(uap, new), SCARG(uap, newlen), p);
244 	if (SCARG(uap, old) != NULL) {
245 		uvm_vsunlock(p, SCARG(uap, old), savelen);
246 		(void) lockmgr(&sysctl_memlock, LK_RELEASE, NULL);
247 	}
248 	if (error)
249 		return (error);
250 	if (SCARG(uap, oldlenp))
251 		error = copyout(&oldlen, SCARG(uap, oldlenp), sizeof(oldlen));
252 	return (error);
253 }
254 
255 /*
256  * Attributes stored in the kernel.
257  */
258 char hostname[MAXHOSTNAMELEN];
259 int hostnamelen;
260 
261 char domainname[MAXHOSTNAMELEN];
262 int domainnamelen;
263 
264 long hostid;
265 
266 #ifdef INSECURE
267 int securelevel = -1;
268 #else
269 int securelevel = 0;
270 #endif
271 
272 #ifndef DEFCORENAME
273 #define	DEFCORENAME	"%n.core"
274 #endif
275 char defcorename[MAXPATHLEN] = DEFCORENAME;
276 int defcorenamelen = sizeof(DEFCORENAME);
277 
278 extern	int	kern_logsigexit;
279 extern	fixpt_t	ccpu;
280 extern	int	forkfsleep;
281 extern	int	dumponpanic;
282 
283 #ifndef MULTIPROCESSOR
284 #define sysctl_ncpus() 1
285 #endif
286 
287 #ifdef MULTIPROCESSOR
288 
289 #ifndef CPU_INFO_FOREACH
290 #define CPU_INFO_ITERATOR int
291 #define CPU_INFO_FOREACH(cii, ci) cii = 0, ci = curcpu(); ci != NULL; ci = NULL
292 #endif
293 
294 static int
295 sysctl_docptime(void *oldp, size_t *oldlenp, void *newp)
296 {
297 	u_int64_t cp_time[CPUSTATES];
298 	int i;
299 	struct cpu_info *ci;
300 	CPU_INFO_ITERATOR cii;
301 
302 	for (i = 0; i < CPUSTATES; i++)
303 		cp_time[i] = 0;
304 
305 	for (CPU_INFO_FOREACH(cii, ci)) {
306 		for (i = 0; i < CPUSTATES; i++)
307 			cp_time[i] += ci->ci_schedstate.spc_cp_time[i];
308 	}
309 	return (sysctl_rdstruct(oldp, oldlenp, newp,
310 	    cp_time, sizeof(cp_time)));
311 }
312 
313 static int
314 sysctl_ncpus(void)
315 {
316 	struct cpu_info *ci;
317 	CPU_INFO_ITERATOR cii;
318 
319 	int ncpus = 0;
320 	for (CPU_INFO_FOREACH(cii, ci))
321 		ncpus++;
322 	return (ncpus);
323 }
324 
325 #endif
326 
327 /*
328  * kernel related system variables.
329  */
330 int
331 kern_sysctl(int *name, u_int namelen, void *oldp, size_t *oldlenp,
332     void *newp, size_t newlen, struct proc *p)
333 {
334 	int error, level, inthostid;
335 	int old_autonicetime;
336 	int old_vnodes;
337 	dev_t consdev;
338 #if NRND > 0
339 	int v;
340 #endif
341 
342 	/* All sysctl names at this level, except for a few, are terminal. */
343 	switch (name[0]) {
344 	case KERN_PROC:
345 	case KERN_PROC2:
346 	case KERN_LWP:
347 	case KERN_PROF:
348 	case KERN_MBUF:
349 	case KERN_PROC_ARGS:
350 	case KERN_SYSVIPC_INFO:
351 	case KERN_PIPE:
352 	case KERN_TKSTAT:
353 		/* Not terminal. */
354 		break;
355 	default:
356 		if (namelen != 1)
357 			return (ENOTDIR);	/* overloaded */
358 	}
359 
360 	switch (name[0]) {
361 	case KERN_OSTYPE:
362 		return (sysctl_rdstring(oldp, oldlenp, newp, ostype));
363 	case KERN_OSRELEASE:
364 		return (sysctl_rdstring(oldp, oldlenp, newp, osrelease));
365 	case KERN_OSREV:
366 		return (sysctl_rdint(oldp, oldlenp, newp, __NetBSD_Version__));
367 	case KERN_VERSION:
368 		return (sysctl_rdstring(oldp, oldlenp, newp, version));
369 	case KERN_MAXVNODES:
370 		old_vnodes = desiredvnodes;
371 		error = sysctl_int(oldp, oldlenp, newp, newlen, &desiredvnodes);
372 		if (newp && !error) {
373 			if (old_vnodes > desiredvnodes) {
374 				error = vfs_drainvnodes(desiredvnodes, p);
375 				if (error) {
376 					desiredvnodes = old_vnodes;
377 					return error;
378 				}
379 			}
380 			vfs_reinit();
381 			nchreinit();
382 		}
383 		return (error);
384 	case KERN_MAXPROC:
385 	    {
386 		int nmaxproc = maxproc;
387 
388 		error = sysctl_int(oldp, oldlenp, newp, newlen, &nmaxproc);
389 
390 		if (!error && newp) {
391 			if (nmaxproc < 0 || nmaxproc >= PID_MAX)
392 				return (EINVAL);
393 
394 #ifdef __HAVE_CPU_MAXPROC
395 			if (nmaxproc > cpu_maxproc())
396 				return (EINVAL);
397 #endif
398 			maxproc = nmaxproc;
399 		}
400 
401 		return (error);
402 	    }
403 	case KERN_MAXFILES:
404 		return (sysctl_int(oldp, oldlenp, newp, newlen, &maxfiles));
405 	case KERN_ARGMAX:
406 		return (sysctl_rdint(oldp, oldlenp, newp, ARG_MAX));
407 	case KERN_SECURELVL:
408 		level = securelevel;
409 		if ((error = sysctl_int(oldp, oldlenp, newp, newlen, &level)) ||
410 		    newp == NULL)
411 			return (error);
412 		if (level < securelevel && p->p_pid != 1)
413 			return (EPERM);
414 		securelevel = level;
415 		return (0);
416 	case KERN_HOSTNAME:
417 		error = sysctl_string(oldp, oldlenp, newp, newlen,
418 		    hostname, sizeof(hostname));
419 		if (newp && !error)
420 			hostnamelen = newlen;
421 		return (error);
422 	case KERN_DOMAINNAME:
423 		error = sysctl_string(oldp, oldlenp, newp, newlen,
424 		    domainname, sizeof(domainname));
425 		if (newp && !error)
426 			domainnamelen = newlen;
427 		return (error);
428 	case KERN_HOSTID:
429 		inthostid = hostid;  /* XXX assumes sizeof long <= sizeof int */
430 		error = sysctl_int(oldp, oldlenp, newp, newlen, &inthostid);
431 		if (newp && !error)
432 			hostid = inthostid;
433 		return (error);
434 	case KERN_CLOCKRATE:
435 		return (sysctl_clockrate(oldp, oldlenp));
436 	case KERN_BOOTTIME:
437 		return (sysctl_rdstruct(oldp, oldlenp, newp, &boottime,
438 		    sizeof(struct timeval)));
439 	case KERN_VNODE:
440 		return (sysctl_vnode(oldp, oldlenp, p));
441 	case KERN_PROC:
442 	case KERN_PROC2:
443 		return (sysctl_doeproc(name, namelen, oldp, oldlenp));
444 	case KERN_LWP:
445 		return (sysctl_dolwp(name, namelen, oldp, oldlenp));
446 	case KERN_PROC_ARGS:
447 		return (sysctl_procargs(name + 1, namelen - 1,
448 		    oldp, oldlenp, p));
449 	case KERN_FILE:
450 		return (sysctl_file(oldp, oldlenp));
451 #ifdef GPROF
452 	case KERN_PROF:
453 		return (sysctl_doprof(name + 1, namelen - 1, oldp, oldlenp,
454 		    newp, newlen));
455 #endif
456 	case KERN_POSIX1:
457 		return (sysctl_rdint(oldp, oldlenp, newp, _POSIX_VERSION));
458 	case KERN_NGROUPS:
459 		return (sysctl_rdint(oldp, oldlenp, newp, NGROUPS_MAX));
460 	case KERN_JOB_CONTROL:
461 		return (sysctl_rdint(oldp, oldlenp, newp, 1));
462 	case KERN_SAVED_IDS:
463 #ifdef _POSIX_SAVED_IDS
464 		return (sysctl_rdint(oldp, oldlenp, newp, 1));
465 #else
466 		return (sysctl_rdint(oldp, oldlenp, newp, 0));
467 #endif
468 	case KERN_MAXPARTITIONS:
469 		return (sysctl_rdint(oldp, oldlenp, newp, MAXPARTITIONS));
470 	case KERN_RAWPARTITION:
471 		return (sysctl_rdint(oldp, oldlenp, newp, RAW_PART));
472 #ifdef NTP
473 	case KERN_NTPTIME:
474 		return (sysctl_ntptime(oldp, oldlenp));
475 #endif
476 	case KERN_AUTONICETIME:
477 		old_autonicetime = autonicetime;
478 		error = sysctl_int(oldp, oldlenp, newp, newlen, &autonicetime);
479 		if (autonicetime < 0)
480  			autonicetime = old_autonicetime;
481 		return (error);
482 	case KERN_AUTONICEVAL:
483 		error = sysctl_int(oldp, oldlenp, newp, newlen, &autoniceval);
484 		if (autoniceval < PRIO_MIN)
485 			autoniceval = PRIO_MIN;
486 		if (autoniceval > PRIO_MAX)
487 			autoniceval = PRIO_MAX;
488 		return (error);
489 	case KERN_RTC_OFFSET:
490 		return (sysctl_rdint(oldp, oldlenp, newp, rtc_offset));
491 	case KERN_ROOT_DEVICE:
492 		return (sysctl_rdstring(oldp, oldlenp, newp,
493 		    root_device->dv_xname));
494 	case KERN_MSGBUFSIZE:
495 		/*
496 		 * deal with cases where the message buffer has
497 		 * become corrupted.
498 		 */
499 		if (!msgbufenabled || msgbufp->msg_magic != MSG_MAGIC) {
500 			msgbufenabled = 0;
501 			return (ENXIO);
502 		}
503 		return (sysctl_rdint(oldp, oldlenp, newp, msgbufp->msg_bufs));
504 	case KERN_FSYNC:
505 		return (sysctl_rdint(oldp, oldlenp, newp, 1));
506 	case KERN_SYSVMSG:
507 #ifdef SYSVMSG
508 		return (sysctl_rdint(oldp, oldlenp, newp, 1));
509 #else
510 		return (sysctl_rdint(oldp, oldlenp, newp, 0));
511 #endif
512 	case KERN_SYSVSEM:
513 #ifdef SYSVSEM
514 		return (sysctl_rdint(oldp, oldlenp, newp, 1));
515 #else
516 		return (sysctl_rdint(oldp, oldlenp, newp, 0));
517 #endif
518 	case KERN_SYSVSHM:
519 #ifdef SYSVSHM
520 		return (sysctl_rdint(oldp, oldlenp, newp, 1));
521 #else
522 		return (sysctl_rdint(oldp, oldlenp, newp, 0));
523 #endif
524  	case KERN_DEFCORENAME:
525 		if (newp && newlen < 1)
526 			return (EINVAL);
527 		error = sysctl_string(oldp, oldlenp, newp, newlen,
528 		    defcorename, sizeof(defcorename));
529 		if (newp && !error)
530 			defcorenamelen = newlen;
531 		return (error);
532 	case KERN_SYNCHRONIZED_IO:
533 		return (sysctl_rdint(oldp, oldlenp, newp, 1));
534 	case KERN_IOV_MAX:
535 		return (sysctl_rdint(oldp, oldlenp, newp, IOV_MAX));
536 	case KERN_MBUF:
537 		return (sysctl_dombuf(name + 1, namelen - 1, oldp, oldlenp,
538 		    newp, newlen));
539 	case KERN_MAPPED_FILES:
540 		return (sysctl_rdint(oldp, oldlenp, newp, 1));
541 	case KERN_MEMLOCK:
542 		return (sysctl_rdint(oldp, oldlenp, newp, 1));
543 	case KERN_MEMLOCK_RANGE:
544 		return (sysctl_rdint(oldp, oldlenp, newp, 1));
545 	case KERN_MEMORY_PROTECTION:
546 		return (sysctl_rdint(oldp, oldlenp, newp, 1));
547 	case KERN_LOGIN_NAME_MAX:
548 		return (sysctl_rdint(oldp, oldlenp, newp, LOGIN_NAME_MAX));
549 	case KERN_LOGSIGEXIT:
550 		return (sysctl_int(oldp, oldlenp, newp, newlen,
551 		    &kern_logsigexit));
552 	case KERN_FSCALE:
553 		return (sysctl_rdint(oldp, oldlenp, newp, FSCALE));
554 	case KERN_CCPU:
555 		return (sysctl_rdint(oldp, oldlenp, newp, ccpu));
556 	case KERN_CP_TIME:
557 #ifndef MULTIPROCESSOR
558 		return (sysctl_rdstruct(oldp, oldlenp, newp,
559 		    curcpu()->ci_schedstate.spc_cp_time,
560 		    sizeof(curcpu()->ci_schedstate.spc_cp_time)));
561 #else
562 		return (sysctl_docptime(oldp, oldlenp, newp));
563 #endif
564 #if defined(SYSVMSG) || defined(SYSVSEM) || defined(SYSVSHM)
565 	case KERN_SYSVIPC_INFO:
566 		return (sysctl_sysvipc(name + 1, namelen - 1, oldp, oldlenp));
567 #endif
568 	case KERN_MSGBUF:
569 		return (sysctl_msgbuf(oldp, oldlenp));
570 	case KERN_CONSDEV:
571 		if (cn_tab != NULL)
572 			consdev = cn_tab->cn_dev;
573 		else
574 			consdev = NODEV;
575 		return (sysctl_rdstruct(oldp, oldlenp, newp, &consdev,
576 		    sizeof consdev));
577 #if NPTY > 0
578 	case KERN_MAXPTYS:
579 		return (sysctl_pty(oldp, oldlenp, newp, newlen));
580 #endif
581 #ifndef PIPE_SOCKETPAIR
582 	case KERN_PIPE:
583 		return (sysctl_dopipe(name + 1, namelen - 1, oldp, oldlenp,
584 		    newp, newlen));
585 #endif
586 	case KERN_MAXPHYS:
587 		return (sysctl_rdint(oldp, oldlenp, newp, MAXPHYS));
588 	case KERN_SOMAXKVA:
589 	    {
590 		int new_somaxkva = somaxkva;
591 
592 		error = sysctl_int(oldp, oldlenp, newp, newlen, &new_somaxkva);
593 		if (newp && !error) {
594 			if (new_somaxkva < (16 * 1024 * 1024)) /* sanity */
595 				return (EINVAL);
596 			somaxkva = new_somaxkva;
597 		}
598 		return (error);
599 	    }
600 	case KERN_SBMAX:
601 	    {
602 		int new_sbmax = sb_max;
603 
604 		error = sysctl_int(oldp, oldlenp, newp, newlen, &new_sbmax);
605 		if (newp && !error) {
606 			if (new_sbmax < (16 * 1024)) /* sanity */
607 				return (EINVAL);
608 			sb_max = new_sbmax;
609 		}
610 		return (error);
611 	    }
612 	case KERN_TKSTAT:
613 		return (sysctl_dotkstat(name + 1, namelen - 1, oldp, oldlenp,
614 		    newp));
615 	case KERN_MONOTONIC_CLOCK:	/* XXX _POSIX_VERSION */
616 		return (sysctl_rdint(oldp, oldlenp, newp, 200112));
617 	case KERN_URND:
618 #if NRND > 0
619 		if (rnd_extract_data(&v, sizeof(v), RND_EXTRACT_ANY) ==
620 		    sizeof(v))
621 			return (sysctl_rdint(oldp, oldlenp, newp, v));
622 		else
623 			return (EIO);	/*XXX*/
624 #else
625 		return (EOPNOTSUPP);
626 #endif
627 	case KERN_LABELSECTOR:
628 		return (sysctl_rdint(oldp, oldlenp, newp, LABELSECTOR));
629 	case KERN_LABELOFFSET:
630 		return (sysctl_rdint(oldp, oldlenp, newp, LABELOFFSET));
631 	case KERN_FORKFSLEEP:
632 	    {
633 		/* userland sees value in ms, internally is in ticks */
634 		int timo, lsleep = forkfsleep * 1000 / hz;
635 
636 		error = sysctl_int(oldp, oldlenp, newp, newlen, &lsleep);
637 		if (newp && !error) {
638 			/* refuse negative values, and overly 'long time' */
639 			if (lsleep < 0 || lsleep > MAXSLP * 1000)
640 				return (EINVAL);
641 
642 			timo = mstohz(lsleep);
643 
644 			/* if the interval is >0 ms && <1 tick, use 1 tick */
645 			if (lsleep != 0 && timo == 0)
646 				forkfsleep = 1;
647 			else
648 				forkfsleep = timo;
649 		}
650 		return (error);
651 	    }
652 	case KERN_POSIX_THREADS:	/* XXX _POSIX_VERSION */
653 		return (sysctl_rdint(oldp, oldlenp, newp, 200112));
654 	case KERN_POSIX_SEMAPHORES:	/* XXX _POSIX_VERSION */
655 #ifdef P1003_1B_SEMAPHORE
656 		return (sysctl_rdint(oldp, oldlenp, newp, 200112));
657 #else
658 		return (sysctl_rdint(oldp, oldlenp, newp, 0));
659 #endif
660 	case KERN_POSIX_BARRIERS:	/* XXX _POSIX_VERSION */
661 		return (sysctl_rdint(oldp, oldlenp, newp, 200112));
662 	case KERN_POSIX_TIMERS:		/* XXX _POSIX_VERSION */
663 		return (sysctl_rdint(oldp, oldlenp, newp, 200112));
664 	case KERN_POSIX_SPIN_LOCKS:	/* XXX _POSIX_VERSION */
665 		return (sysctl_rdint(oldp, oldlenp, newp, 200112));
666 	case KERN_POSIX_READER_WRITER_LOCKS:	/* XXX _POSIX_VERSION */
667 		return (sysctl_rdint(oldp, oldlenp, newp, 200112));
668 	case KERN_DUMP_ON_PANIC:
669 		return (sysctl_int(oldp, oldlenp, newp, newlen, &dumponpanic));
670 
671 	default:
672 		return (EOPNOTSUPP);
673 	}
674 	/* NOTREACHED */
675 }
676 
677 /*
678  * hardware related system variables.
679  */
680 int
681 hw_sysctl(int *name, u_int namelen, void *oldp, size_t *oldlenp,
682     void *newp, size_t newlen, struct proc *p)
683 {
684 
685 	/* All sysctl names at this level, except for a few, are terminal. */
686 	switch (name[0]) {
687 	case HW_DISKSTATS:
688 		/* Not terminal. */
689 		break;
690 	default:
691 		if (namelen != 1)
692 			return (ENOTDIR);	/* overloaded */
693 	}
694 
695 	switch (name[0]) {
696 	case HW_MACHINE:
697 		return (sysctl_rdstring(oldp, oldlenp, newp, machine));
698 	case HW_MACHINE_ARCH:
699 		return (sysctl_rdstring(oldp, oldlenp, newp, machine_arch));
700 	case HW_MODEL:
701 		return (sysctl_rdstring(oldp, oldlenp, newp, cpu_model));
702 	case HW_NCPU:
703 		return (sysctl_rdint(oldp, oldlenp, newp, sysctl_ncpus()));
704 	case HW_BYTEORDER:
705 		return (sysctl_rdint(oldp, oldlenp, newp, BYTE_ORDER));
706 	case HW_PHYSMEM:
707 	    {
708 		u_int rval;
709 
710 		if ((u_int)physmem > (UINT_MAX / PAGE_SIZE))
711 			rval = UINT_MAX;
712 		else
713 			rval = physmem * PAGE_SIZE;
714 		return (sysctl_rdint(oldp, oldlenp, newp, rval));
715 	    }
716 	case HW_PHYSMEM64:
717 		return (sysctl_rdquad(oldp, oldlenp, newp,
718 		    (u_quad_t)physmem * PAGE_SIZE));
719 	case HW_USERMEM:
720 	    {
721 		u_int rval;
722 
723 		if ((u_int)(physmem - uvmexp.wired) > (UINT_MAX / PAGE_SIZE))
724 			rval = UINT_MAX;
725 		else
726 			rval = (physmem - uvmexp.wired) * PAGE_SIZE;
727 		return (sysctl_rdint(oldp, oldlenp, newp, rval));
728 	    }
729 	case HW_USERMEM64:
730 		return (sysctl_rdquad(oldp, oldlenp, newp,
731 		    (u_quad_t)(physmem - uvmexp.wired) * PAGE_SIZE));
732 	case HW_PAGESIZE:
733 		return (sysctl_rdint(oldp, oldlenp, newp, PAGE_SIZE));
734 	case HW_ALIGNBYTES:
735 		return (sysctl_rdint(oldp, oldlenp, newp, ALIGNBYTES));
736 	case HW_DISKNAMES:
737 		return (sysctl_disknames(oldp, oldlenp));
738 	case HW_DISKSTATS:
739 		return (sysctl_diskstats(name + 1, namelen - 1, oldp, oldlenp));
740 	case HW_CNMAGIC: {
741 		char magic[CNS_LEN];
742 		int error;
743 
744 		if (oldp)
745 			cn_get_magic(magic, CNS_LEN);
746 		error = sysctl_string(oldp, oldlenp, newp, newlen,
747 		    magic, sizeof(magic));
748 		if (newp && !error) {
749 			error = cn_set_magic(magic);
750 		}
751 		return (error);
752 	}
753 	default:
754 		return (EOPNOTSUPP);
755 	}
756 	/* NOTREACHED */
757 }
758 
759 #ifdef DEBUG
760 /*
761  * Debugging related system variables.
762  */
763 struct ctldebug /* debug0, */ /* debug1, */ debug2, debug3, debug4;
764 struct ctldebug debug5, debug6, debug7, debug8, debug9;
765 struct ctldebug debug10, debug11, debug12, debug13, debug14;
766 struct ctldebug debug15, debug16, debug17, debug18, debug19;
767 static struct ctldebug *debugvars[CTL_DEBUG_MAXID] = {
768 	&debug0, &debug1, &debug2, &debug3, &debug4,
769 	&debug5, &debug6, &debug7, &debug8, &debug9,
770 	&debug10, &debug11, &debug12, &debug13, &debug14,
771 	&debug15, &debug16, &debug17, &debug18, &debug19,
772 };
773 
774 int
775 debug_sysctl(int *name, u_int namelen, void *oldp, size_t *oldlenp,
776     void *newp, size_t newlen, struct proc *p)
777 {
778 	struct ctldebug *cdp;
779 
780 	/* all sysctl names at this level are name and field */
781 	if (namelen != 2)
782 		return (ENOTDIR);		/* overloaded */
783 	if (name[0] >= CTL_DEBUG_MAXID)
784 		return (EOPNOTSUPP);
785 	cdp = debugvars[name[0]];
786 	if (cdp->debugname == 0)
787 		return (EOPNOTSUPP);
788 	switch (name[1]) {
789 	case CTL_DEBUG_NAME:
790 		return (sysctl_rdstring(oldp, oldlenp, newp, cdp->debugname));
791 	case CTL_DEBUG_VALUE:
792 		return (sysctl_int(oldp, oldlenp, newp, newlen, cdp->debugvar));
793 	default:
794 		return (EOPNOTSUPP);
795 	}
796 	/* NOTREACHED */
797 }
798 #endif /* DEBUG */
799 
800 int
801 proc_sysctl(int *name, u_int namelen, void *oldp, size_t *oldlenp,
802     void *newp, size_t newlen, struct proc *p)
803 {
804 	struct proc *ptmp = NULL;
805 	const struct proclist_desc *pd;
806 	int error = 0;
807 	struct rlimit alim;
808 	struct plimit *newplim;
809 	char *tmps = NULL;
810 	size_t len, curlen;
811 	u_int i;
812 
813 	if (namelen < 2)
814 		return (EINVAL);
815 
816 	if (name[0] == PROC_CURPROC) {
817 		ptmp = p;
818 	} else {
819 		proclist_lock_read();
820 		for (pd = proclists; pd->pd_list != NULL; pd++) {
821 			for (ptmp = LIST_FIRST(pd->pd_list); ptmp != NULL;
822 			    ptmp = LIST_NEXT(ptmp, p_list)) {
823 				/* Skip embryonic processes. */
824 				if (ptmp->p_stat == SIDL)
825 					continue;
826 				if (ptmp->p_pid == (pid_t)name[0])
827 					break;
828 			}
829 			if (ptmp != NULL)
830 				break;
831 		}
832 		proclist_unlock_read();
833 		if (ptmp == NULL)
834 			return (ESRCH);
835 		if (p->p_ucred->cr_uid != 0) {
836 			if (p->p_cred->p_ruid != ptmp->p_cred->p_ruid ||
837 			    p->p_cred->p_ruid != ptmp->p_cred->p_svuid)
838 				return (EPERM);
839 			if (ptmp->p_cred->p_rgid != ptmp->p_cred->p_svgid)
840 				return (EPERM); /* sgid proc */
841 			for (i = 0; i < p->p_ucred->cr_ngroups; i++) {
842 				if (p->p_ucred->cr_groups[i] ==
843 				    ptmp->p_cred->p_rgid)
844 					break;
845 			}
846 			if (i == p->p_ucred->cr_ngroups)
847 				return (EPERM);
848 		}
849 	}
850 	switch (name[1]) {
851 	case PROC_PID_STOPFORK:
852 		if (namelen != 2)
853 			return (EINVAL);
854 		i = ((ptmp->p_flag & P_STOPFORK) != 0);
855 		if ((error = sysctl_int(oldp, oldlenp, newp, newlen, &i)) != 0)
856 			return (error);
857 		if (i != 0)
858 			ptmp->p_flag |= P_STOPFORK;
859 		else
860 			ptmp->p_flag &= ~P_STOPFORK;
861 		return (0);
862 		break;
863 
864 	case PROC_PID_STOPEXEC:
865 		if (namelen != 2)
866 			return (EINVAL);
867 		i = ((ptmp->p_flag & P_STOPEXEC) != 0);
868 		if ((error = sysctl_int(oldp, oldlenp, newp, newlen, &i)) != 0)
869 			return (error);
870 		if (i != 0)
871 			ptmp->p_flag |= P_STOPEXEC;
872 		else
873 			ptmp->p_flag &= ~P_STOPEXEC;
874 		return (0);
875 		break;
876 
877 	case PROC_PID_CORENAME:
878 		if (namelen != 2)
879 			return (EINVAL);
880 		/*
881 		 * Can't use sysctl_string() here because we may malloc a new
882 		 * area during the process, so we have to do it by hand.
883 		 */
884 		curlen = strlen(ptmp->p_limit->pl_corename) + 1;
885 		if (oldlenp && *oldlenp < curlen) {
886 			if (!oldp)
887 				*oldlenp = curlen;
888 			return (ENOMEM);
889 		}
890 		if (newp) {
891 			if (securelevel > 2)
892 				return (EPERM);
893 			if (newlen > MAXPATHLEN)
894 				return (ENAMETOOLONG);
895 			tmps = malloc(newlen + 1, M_TEMP, M_WAITOK);
896 			if (tmps == NULL)
897 				return (ENOMEM);
898 			error = copyin(newp, tmps, newlen + 1);
899 			tmps[newlen] = '\0';
900 			if (error)
901 				goto cleanup;
902 			/* Enforce to be either 'core' for end with '.core' */
903 			if (newlen < 4) {	/* c.o.r.e */
904 				error = EINVAL;
905 				goto cleanup;
906 			}
907 			len = newlen - 4;
908 			if (len > 0) {
909 				if (tmps[len - 1] != '.' &&
910 				    tmps[len - 1] != '/') {
911 					error = EINVAL;
912 					goto cleanup;
913 				}
914 			}
915 			if (strcmp(&tmps[len], "core") != 0) {
916 				error = EINVAL;
917 				goto cleanup;
918 			}
919 		}
920 		if (oldp && oldlenp) {
921 			*oldlenp = curlen;
922 			error = copyout(ptmp->p_limit->pl_corename, oldp,
923 			    curlen);
924 		}
925 		if (newp && error == 0) {
926 			/* if the 2 strings are identical, don't limcopy() */
927 			if (strcmp(tmps, ptmp->p_limit->pl_corename) == 0) {
928 				error = 0;
929 				goto cleanup;
930 			}
931 			if (ptmp->p_limit->p_refcnt > 1 &&
932 			    (ptmp->p_limit->p_lflags & PL_SHAREMOD) == 0) {
933 				newplim = limcopy(ptmp->p_limit);
934 				limfree(ptmp->p_limit);
935 				ptmp->p_limit = newplim;
936 			}
937 			if (ptmp->p_limit->pl_corename != defcorename) {
938 				free(ptmp->p_limit->pl_corename, M_TEMP);
939 			}
940 			ptmp->p_limit->pl_corename = tmps;
941 			return (0);
942 		}
943 cleanup:
944 		if (tmps)
945 			free(tmps, M_TEMP);
946 		return (error);
947 		break;
948 
949 	case PROC_PID_LIMIT:
950 		if (namelen != 4 || name[2] >= PROC_PID_LIMIT_MAXID)
951 			return (EINVAL);
952 		memcpy(&alim, &ptmp->p_rlimit[name[2] - 1], sizeof(alim));
953 		if (name[3] == PROC_PID_LIMIT_TYPE_HARD)
954 			error = sysctl_quad(oldp, oldlenp, newp, newlen,
955 			    &alim.rlim_max);
956 		else if (name[3] == PROC_PID_LIMIT_TYPE_SOFT)
957 			error = sysctl_quad(oldp, oldlenp, newp, newlen,
958 			    &alim.rlim_cur);
959 		else
960 			error = (EINVAL);
961 
962 		if (error)
963 			return (error);
964 
965 		if (newp)
966 			error = dosetrlimit(ptmp, p->p_cred,
967 			    name[2] - 1, &alim);
968 		return (error);
969 		break;
970 
971 	default:
972 		return (EINVAL);
973 		break;
974 	}
975 	/* NOTREACHED */
976 	return (EINVAL);
977 }
978 
979 int
980 emul_sysctl(int *name, u_int namelen, void *oldp, size_t *oldlenp,
981     void *newp, size_t newlen, struct proc *p)
982 {
983 	static struct {
984 		const char *name;
985 		int type;
986 	} emulations[] = CTL_EMUL_NAMES;
987 	const struct emul *e;
988 	const char *ename;
989 #ifdef LKM
990 	extern struct lock exec_lock;	/* XXX */
991 	int error;
992 #else
993 	extern int nexecs_builtin;
994 	extern const struct execsw execsw_builtin[];
995 	int i;
996 #endif
997 
998 	/* all sysctl names at this level are name and field */
999 	if (namelen < 2)
1000 		return (ENOTDIR);		/* overloaded */
1001 
1002 	if ((u_int) name[0] >= EMUL_MAXID || name[0] == 0)
1003 		return (EOPNOTSUPP);
1004 
1005 	ename = emulations[name[0]].name;
1006 
1007 #ifdef LKM
1008 	lockmgr(&exec_lock, LK_SHARED, NULL);
1009 	if ((e = emul_search(ename))) {
1010 		error = (*e->e_sysctl)(name + 1, namelen - 1, oldp, oldlenp,
1011 		    newp, newlen, p);
1012 	} else
1013 		error = EOPNOTSUPP;
1014 	lockmgr(&exec_lock, LK_RELEASE, NULL);
1015 
1016 	return (error);
1017 #else
1018 	for (i = 0; i < nexecs_builtin; i++) {
1019 		e = execsw_builtin[i].es_emul;
1020 		/*
1021 		 * In order to match e.g. e->e_name "irix o32"
1022 		 * with ename "irix", we limit the comparison
1023 		 * to the length of ename.
1024 		 */
1025 		if (e == NULL ||
1026 		    strncmp(ename, e->e_name, strlen(ename)) != 0 ||
1027 		    e->e_sysctl == NULL)
1028 			continue;
1029 
1030 		return ((*e->e_sysctl)(name + 1, namelen - 1, oldp, oldlenp,
1031 		    newp, newlen, p));
1032 	}
1033 
1034 	return (EOPNOTSUPP);
1035 #endif
1036 }
1037 /*
1038  * Convenience macros.
1039  */
1040 
1041 #define SYSCTL_SCALAR_CORE_LEN(oldp, oldlenp, valp, len) 		\
1042 	if (oldlenp) {							\
1043 		if (!oldp)						\
1044 			*oldlenp = len;					\
1045 		else {							\
1046 			if (*oldlenp < len)				\
1047 				return (ENOMEM);			\
1048 			*oldlenp = len;					\
1049 			error = copyout((caddr_t)valp, oldp, len);	\
1050 		}							\
1051 	}
1052 
1053 #define SYSCTL_SCALAR_CORE_TYP(oldp, oldlenp, valp, typ) \
1054 	SYSCTL_SCALAR_CORE_LEN(oldp, oldlenp, valp, sizeof(typ))
1055 
1056 #define SYSCTL_SCALAR_NEWPCHECK_LEN(newp, newlen, len)	\
1057 	if (newp && newlen != len)			\
1058 		return (EINVAL);
1059 
1060 #define SYSCTL_SCALAR_NEWPCHECK_TYP(newp, newlen, typ)	\
1061 	SYSCTL_SCALAR_NEWPCHECK_LEN(newp, newlen, sizeof(typ))
1062 
1063 #define SYSCTL_SCALAR_NEWPCOP_LEN(newp, valp, len)	\
1064 	if (error == 0 && newp)				\
1065 		error = copyin(newp, valp, len);
1066 
1067 #define SYSCTL_SCALAR_NEWPCOP_TYP(newp, valp, typ)	\
1068 	SYSCTL_SCALAR_NEWPCOP_LEN(newp, valp, sizeof(typ))
1069 
1070 #define SYSCTL_STRING_CORE(oldp, oldlenp, str)		\
1071 	if (oldlenp) {					\
1072 		len = strlen(str) + 1;			\
1073 		if (!oldp)				\
1074 			*oldlenp = len;			\
1075 		else {					\
1076 			if (*oldlenp < len) {		\
1077 				err2 = ENOMEM;		\
1078 				len = *oldlenp;		\
1079 			} else				\
1080 				*oldlenp = len;		\
1081 			error = copyout(str, oldp, len);\
1082 			if (error == 0)			\
1083 				error = err2;		\
1084 		}					\
1085 	}
1086 
1087 /*
1088  * Validate parameters and get old / set new parameters
1089  * for an integer-valued sysctl function.
1090  */
1091 int
1092 sysctl_int(void *oldp, size_t *oldlenp, void *newp, size_t newlen, int *valp)
1093 {
1094 	int error = 0;
1095 
1096 	SYSCTL_SCALAR_NEWPCHECK_TYP(newp, newlen, int)
1097 	SYSCTL_SCALAR_CORE_TYP(oldp, oldlenp, valp, int)
1098 	SYSCTL_SCALAR_NEWPCOP_TYP(newp, valp, int)
1099 
1100 	return (error);
1101 }
1102 
1103 
1104 /*
1105  * As above, but read-only.
1106  */
1107 int
1108 sysctl_rdint(void *oldp, size_t *oldlenp, void *newp, int val)
1109 {
1110 	int error = 0;
1111 
1112 	if (newp)
1113 		return (EPERM);
1114 
1115 	SYSCTL_SCALAR_CORE_TYP(oldp, oldlenp, &val, int)
1116 
1117 	return (error);
1118 }
1119 
1120 /*
1121  * Validate parameters and get old / set new parameters
1122  * for an quad-valued sysctl function.
1123  */
1124 int
1125 sysctl_quad(void *oldp, size_t *oldlenp, void *newp, size_t newlen,
1126     quad_t *valp)
1127 {
1128 	int error = 0;
1129 
1130 	SYSCTL_SCALAR_NEWPCHECK_TYP(newp, newlen, quad_t)
1131 	SYSCTL_SCALAR_CORE_TYP(oldp, oldlenp, valp, quad_t)
1132 	SYSCTL_SCALAR_NEWPCOP_TYP(newp, valp, quad_t)
1133 
1134 	return (error);
1135 }
1136 
1137 /*
1138  * As above, but read-only.
1139  */
1140 int
1141 sysctl_rdquad(void *oldp, size_t *oldlenp, void *newp, quad_t val)
1142 {
1143 	int error = 0;
1144 
1145 	if (newp)
1146 		return (EPERM);
1147 
1148 	SYSCTL_SCALAR_CORE_TYP(oldp, oldlenp, &val, quad_t)
1149 
1150 	return (error);
1151 }
1152 
1153 /*
1154  * Validate parameters and get old / set new parameters
1155  * for a string-valued sysctl function.
1156  */
1157 int
1158 sysctl_string(void *oldp, size_t *oldlenp, void *newp, size_t newlen, char *str,
1159     size_t maxlen)
1160 {
1161 	int error = 0, err2 = 0;
1162 	size_t len;
1163 
1164 	if (newp && newlen >= maxlen)
1165 		return (EINVAL);
1166 
1167 	SYSCTL_STRING_CORE(oldp, oldlenp, str);
1168 
1169 	if (error == 0 && newp) {
1170 		error = copyin(newp, str, newlen);
1171 		str[newlen] = 0;
1172 	}
1173 	return (error);
1174 }
1175 
1176 /*
1177  * As above, but read-only.
1178  */
1179 int
1180 sysctl_rdstring(void *oldp, size_t *oldlenp, void *newp, const char *str)
1181 {
1182 	int error = 0, err2 = 0;
1183 	size_t len;
1184 
1185 	if (newp)
1186 		return (EPERM);
1187 
1188 	SYSCTL_STRING_CORE(oldp, oldlenp, str);
1189 
1190 	return (error);
1191 }
1192 
1193 /*
1194  * Validate parameters and get old / set new parameters
1195  * for a structure oriented sysctl function.
1196  */
1197 int
1198 sysctl_struct(void *oldp, size_t *oldlenp, void *newp, size_t newlen, void *sp,
1199     size_t len)
1200 {
1201 	int error = 0;
1202 
1203 	SYSCTL_SCALAR_NEWPCHECK_LEN(newp, newlen, len)
1204 	SYSCTL_SCALAR_CORE_LEN(oldp, oldlenp, sp, len)
1205 	SYSCTL_SCALAR_NEWPCOP_LEN(newp, sp, len)
1206 
1207 	return (error);
1208 }
1209 
1210 /*
1211  * Validate parameters and get old parameters
1212  * for a structure oriented sysctl function.
1213  */
1214 int
1215 sysctl_rdstruct(void *oldp, size_t *oldlenp, void *newp, const void *sp,
1216     size_t len)
1217 {
1218 	int error = 0;
1219 
1220 	if (newp)
1221 		return (EPERM);
1222 
1223 	SYSCTL_SCALAR_CORE_LEN(oldp, oldlenp, sp, len)
1224 
1225 	return (error);
1226 }
1227 
1228 /*
1229  * As above, but can return a truncated result.
1230  */
1231 int
1232 sysctl_rdminstruct(void *oldp, size_t *oldlenp, void *newp, const void *sp,
1233     size_t len)
1234 {
1235 	int error = 0;
1236 
1237 	if (newp)
1238 		return (EPERM);
1239 
1240 	len = min(*oldlenp, len);
1241 	SYSCTL_SCALAR_CORE_LEN(oldp, oldlenp, sp, len)
1242 
1243 	return (error);
1244 }
1245 
1246 /*
1247  * Get file structures.
1248  */
1249 static int
1250 sysctl_file(void *vwhere, size_t *sizep)
1251 {
1252 	int error;
1253 	size_t buflen;
1254 	struct file *fp;
1255 	char *start, *where;
1256 
1257 	start = where = vwhere;
1258 	buflen = *sizep;
1259 	if (where == NULL) {
1260 		/*
1261 		 * overestimate by 10 files
1262 		 */
1263 		*sizep = sizeof(filehead) + (nfiles + 10) * sizeof(struct file);
1264 		return (0);
1265 	}
1266 
1267 	/*
1268 	 * first copyout filehead
1269 	 */
1270 	if (buflen < sizeof(filehead)) {
1271 		*sizep = 0;
1272 		return (0);
1273 	}
1274 	error = copyout((caddr_t)&filehead, where, sizeof(filehead));
1275 	if (error)
1276 		return (error);
1277 	buflen -= sizeof(filehead);
1278 	where += sizeof(filehead);
1279 
1280 	/*
1281 	 * followed by an array of file structures
1282 	 */
1283 	LIST_FOREACH(fp, &filehead, f_list) {
1284 		if (buflen < sizeof(struct file)) {
1285 			*sizep = where - start;
1286 			return (ENOMEM);
1287 		}
1288 		error = copyout((caddr_t)fp, where, sizeof(struct file));
1289 		if (error)
1290 			return (error);
1291 		buflen -= sizeof(struct file);
1292 		where += sizeof(struct file);
1293 	}
1294 	*sizep = where - start;
1295 	return (0);
1296 }
1297 
1298 #if defined(SYSVMSG) || defined(SYSVSEM) || defined(SYSVSHM)
1299 #define	FILL_PERM(src, dst) do { \
1300 	(dst)._key = (src)._key; \
1301 	(dst).uid = (src).uid; \
1302 	(dst).gid = (src).gid; \
1303 	(dst).cuid = (src).cuid; \
1304 	(dst).cgid = (src).cgid; \
1305 	(dst).mode = (src).mode; \
1306 	(dst)._seq = (src)._seq; \
1307 } while (/*CONSTCOND*/ 0);
1308 #define	FILL_MSG(src, dst) do { \
1309 	FILL_PERM((src).msg_perm, (dst).msg_perm); \
1310 	(dst).msg_qnum = (src).msg_qnum; \
1311 	(dst).msg_qbytes = (src).msg_qbytes; \
1312 	(dst)._msg_cbytes = (src)._msg_cbytes; \
1313 	(dst).msg_lspid = (src).msg_lspid; \
1314 	(dst).msg_lrpid = (src).msg_lrpid; \
1315 	(dst).msg_stime = (src).msg_stime; \
1316 	(dst).msg_rtime = (src).msg_rtime; \
1317 	(dst).msg_ctime = (src).msg_ctime; \
1318 } while (/*CONSTCOND*/ 0)
1319 #define	FILL_SEM(src, dst) do { \
1320 	FILL_PERM((src).sem_perm, (dst).sem_perm); \
1321 	(dst).sem_nsems = (src).sem_nsems; \
1322 	(dst).sem_otime = (src).sem_otime; \
1323 	(dst).sem_ctime = (src).sem_ctime; \
1324 } while (/*CONSTCOND*/ 0)
1325 #define	FILL_SHM(src, dst) do { \
1326 	FILL_PERM((src).shm_perm, (dst).shm_perm); \
1327 	(dst).shm_segsz = (src).shm_segsz; \
1328 	(dst).shm_lpid = (src).shm_lpid; \
1329 	(dst).shm_cpid = (src).shm_cpid; \
1330 	(dst).shm_atime = (src).shm_atime; \
1331 	(dst).shm_dtime = (src).shm_dtime; \
1332 	(dst).shm_ctime = (src).shm_ctime; \
1333 	(dst).shm_nattch = (src).shm_nattch; \
1334 } while (/*CONSTCOND*/ 0)
1335 
1336 static int
1337 sysctl_sysvipc(int *name, u_int namelen, void *where, size_t *sizep)
1338 {
1339 #ifdef SYSVMSG
1340 	struct msg_sysctl_info *msgsi = NULL;
1341 #endif
1342 #ifdef SYSVSEM
1343 	struct sem_sysctl_info *semsi = NULL;
1344 #endif
1345 #ifdef SYSVSHM
1346 	struct shm_sysctl_info *shmsi = NULL;
1347 #endif
1348 	size_t infosize, dssize, tsize, buflen;
1349 	void *buf = NULL;
1350 	char *start;
1351 	int32_t nds;
1352 	int i, error, ret;
1353 
1354 	if (namelen != 1)
1355 		return (EINVAL);
1356 
1357 	start = where;
1358 	buflen = *sizep;
1359 
1360 	switch (*name) {
1361 	case KERN_SYSVIPC_MSG_INFO:
1362 #ifdef SYSVMSG
1363 		infosize = sizeof(msgsi->msginfo);
1364 		nds = msginfo.msgmni;
1365 		dssize = sizeof(msgsi->msgids[0]);
1366 		break;
1367 #else
1368 		return (EINVAL);
1369 #endif
1370 	case KERN_SYSVIPC_SEM_INFO:
1371 #ifdef SYSVSEM
1372 		infosize = sizeof(semsi->seminfo);
1373 		nds = seminfo.semmni;
1374 		dssize = sizeof(semsi->semids[0]);
1375 		break;
1376 #else
1377 		return (EINVAL);
1378 #endif
1379 	case KERN_SYSVIPC_SHM_INFO:
1380 #ifdef SYSVSHM
1381 		infosize = sizeof(shmsi->shminfo);
1382 		nds = shminfo.shmmni;
1383 		dssize = sizeof(shmsi->shmids[0]);
1384 		break;
1385 #else
1386 		return (EINVAL);
1387 #endif
1388 	default:
1389 		return (EINVAL);
1390 	}
1391 	/*
1392 	 * Round infosize to 64 bit boundary if requesting more than just
1393 	 * the info structure or getting the total data size.
1394 	 */
1395 	if (where == NULL || *sizep > infosize)
1396 		infosize = ((infosize + 7) / 8) * 8;
1397 	tsize = infosize + nds * dssize;
1398 
1399 	/* Return just the total size required. */
1400 	if (where == NULL) {
1401 		*sizep = tsize;
1402 		return (0);
1403 	}
1404 
1405 	/* Not enough room for even the info struct. */
1406 	if (buflen < infosize) {
1407 		*sizep = 0;
1408 		return (ENOMEM);
1409 	}
1410 	buf = malloc(min(tsize, buflen), M_TEMP, M_WAITOK);
1411 	memset(buf, 0, min(tsize, buflen));
1412 
1413 	switch (*name) {
1414 #ifdef SYSVMSG
1415 	case KERN_SYSVIPC_MSG_INFO:
1416 		msgsi = (struct msg_sysctl_info *)buf;
1417 		msgsi->msginfo = msginfo;
1418 		break;
1419 #endif
1420 #ifdef SYSVSEM
1421 	case KERN_SYSVIPC_SEM_INFO:
1422 		semsi = (struct sem_sysctl_info *)buf;
1423 		semsi->seminfo = seminfo;
1424 		break;
1425 #endif
1426 #ifdef SYSVSHM
1427 	case KERN_SYSVIPC_SHM_INFO:
1428 		shmsi = (struct shm_sysctl_info *)buf;
1429 		shmsi->shminfo = shminfo;
1430 		break;
1431 #endif
1432 	}
1433 	buflen -= infosize;
1434 
1435 	ret = 0;
1436 	if (buflen > 0) {
1437 		/* Fill in the IPC data structures.  */
1438 		for (i = 0; i < nds; i++) {
1439 			if (buflen < dssize) {
1440 				ret = ENOMEM;
1441 				break;
1442 			}
1443 			switch (*name) {
1444 #ifdef SYSVMSG
1445 			case KERN_SYSVIPC_MSG_INFO:
1446 				FILL_MSG(msqids[i], msgsi->msgids[i]);
1447 				break;
1448 #endif
1449 #ifdef SYSVSEM
1450 			case KERN_SYSVIPC_SEM_INFO:
1451 				FILL_SEM(sema[i], semsi->semids[i]);
1452 				break;
1453 #endif
1454 #ifdef SYSVSHM
1455 			case KERN_SYSVIPC_SHM_INFO:
1456 				FILL_SHM(shmsegs[i], shmsi->shmids[i]);
1457 				break;
1458 #endif
1459 			}
1460 			buflen -= dssize;
1461 		}
1462 	}
1463 	*sizep -= buflen;
1464 	error = copyout(buf, start, *sizep);
1465 	/* If copyout succeeded, use return code set earlier. */
1466 	if (error == 0)
1467 		error = ret;
1468 	if (buf)
1469 		free(buf, M_TEMP);
1470 	return (error);
1471 }
1472 #endif /* SYSVMSG || SYSVSEM || SYSVSHM */
1473 
1474 static int
1475 sysctl_msgbuf(void *vwhere, size_t *sizep)
1476 {
1477 	char *where = vwhere;
1478 	size_t len, maxlen = *sizep;
1479 	long beg, end;
1480 	int error;
1481 
1482 	/*
1483 	 * deal with cases where the message buffer has
1484 	 * become corrupted.
1485 	 */
1486 	if (!msgbufenabled || msgbufp->msg_magic != MSG_MAGIC) {
1487 		msgbufenabled = 0;
1488 		return (ENXIO);
1489 	}
1490 
1491 	if (where == NULL) {
1492 		/* always return full buffer size */
1493 		*sizep = msgbufp->msg_bufs;
1494 		return (0);
1495 	}
1496 
1497 	error = 0;
1498 	maxlen = min(msgbufp->msg_bufs, maxlen);
1499 
1500 	/*
1501 	 * First, copy from the write pointer to the end of
1502 	 * message buffer.
1503 	 */
1504 	beg = msgbufp->msg_bufx;
1505 	end = msgbufp->msg_bufs;
1506 	while (maxlen > 0) {
1507 		len = min(end - beg, maxlen);
1508 		if (len == 0)
1509 			break;
1510 		error = copyout(&msgbufp->msg_bufc[beg], where, len);
1511 		if (error)
1512 			break;
1513 		where += len;
1514 		maxlen -= len;
1515 
1516 		/*
1517 		 * ... then, copy from the beginning of message buffer to
1518 		 * the write pointer.
1519 		 */
1520 		beg = 0;
1521 		end = msgbufp->msg_bufx;
1522 	}
1523 	return (error);
1524 }
1525 
1526 /*
1527  * try over estimating by 5 procs
1528  */
1529 #define KERN_PROCSLOP	(5 * sizeof(struct kinfo_proc))
1530 
1531 static int
1532 sysctl_doeproc(int *name, u_int namelen, void *vwhere, size_t *sizep)
1533 {
1534 	struct eproc eproc;
1535 	struct kinfo_proc2 kproc2;
1536 	struct kinfo_proc *dp;
1537 	struct proc *p;
1538 	const struct proclist_desc *pd;
1539 	char *where, *dp2;
1540 	int type, op, arg;
1541 	u_int elem_size, elem_count;
1542 	size_t buflen, needed;
1543 	int error;
1544 
1545 	dp = vwhere;
1546 	dp2 = where = vwhere;
1547 	buflen = where != NULL ? *sizep : 0;
1548 	error = 0;
1549 	needed = 0;
1550 	type = name[0];
1551 
1552 	if (type == KERN_PROC) {
1553 		if (namelen != 3 && !(namelen == 2 && name[1] == KERN_PROC_ALL))
1554 			return (EINVAL);
1555 		op = name[1];
1556 		if (op != KERN_PROC_ALL)
1557 			arg = name[2];
1558 		else
1559 			arg = 0;		/* Quell compiler warning */
1560 		elem_size = elem_count = 0;	/* Ditto */
1561 	} else {
1562 		if (namelen != 5)
1563 			return (EINVAL);
1564 		op = name[1];
1565 		arg = name[2];
1566 		elem_size = name[3];
1567 		elem_count = name[4];
1568 	}
1569 
1570 	proclist_lock_read();
1571 
1572 	pd = proclists;
1573 again:
1574 	for (p = LIST_FIRST(pd->pd_list); p != NULL; p = LIST_NEXT(p, p_list)) {
1575 		/*
1576 		 * Skip embryonic processes.
1577 		 */
1578 		if (p->p_stat == SIDL)
1579 			continue;
1580 		/*
1581 		 * TODO - make more efficient (see notes below).
1582 		 * do by session.
1583 		 */
1584 		switch (op) {
1585 
1586 		case KERN_PROC_PID:
1587 			/* could do this with just a lookup */
1588 			if (p->p_pid != (pid_t)arg)
1589 				continue;
1590 			break;
1591 
1592 		case KERN_PROC_PGRP:
1593 			/* could do this by traversing pgrp */
1594 			if (p->p_pgrp->pg_id != (pid_t)arg)
1595 				continue;
1596 			break;
1597 
1598 		case KERN_PROC_SESSION:
1599 			if (p->p_session->s_sid != (pid_t)arg)
1600 				continue;
1601 			break;
1602 
1603 		case KERN_PROC_TTY:
1604 			if (arg == (int) KERN_PROC_TTY_REVOKE) {
1605 				if ((p->p_flag & P_CONTROLT) == 0 ||
1606 				    p->p_session->s_ttyp == NULL ||
1607 				    p->p_session->s_ttyvp != NULL)
1608 					continue;
1609 			} else if ((p->p_flag & P_CONTROLT) == 0 ||
1610 			    p->p_session->s_ttyp == NULL) {
1611 				if ((dev_t)arg != KERN_PROC_TTY_NODEV)
1612 					continue;
1613 			} else if (p->p_session->s_ttyp->t_dev != (dev_t)arg)
1614 				continue;
1615 			break;
1616 
1617 		case KERN_PROC_UID:
1618 			if (p->p_ucred->cr_uid != (uid_t)arg)
1619 				continue;
1620 			break;
1621 
1622 		case KERN_PROC_RUID:
1623 			if (p->p_cred->p_ruid != (uid_t)arg)
1624 				continue;
1625 			break;
1626 
1627 		case KERN_PROC_GID:
1628 			if (p->p_ucred->cr_gid != (uid_t)arg)
1629 				continue;
1630 			break;
1631 
1632 		case KERN_PROC_RGID:
1633 			if (p->p_cred->p_rgid != (uid_t)arg)
1634 				continue;
1635 			break;
1636 
1637 		case KERN_PROC_ALL:
1638 			/* allow everything */
1639 			break;
1640 
1641 		default:
1642 			error = EINVAL;
1643 			goto cleanup;
1644 		}
1645 		if (type == KERN_PROC) {
1646 			if (buflen >= sizeof(struct kinfo_proc)) {
1647 				fill_eproc(p, &eproc);
1648 				error = copyout((caddr_t)p, &dp->kp_proc,
1649 				    sizeof(struct proc));
1650 				if (error)
1651 					goto cleanup;
1652 				error = copyout((caddr_t)&eproc, &dp->kp_eproc,
1653 				    sizeof(eproc));
1654 				if (error)
1655 					goto cleanup;
1656 				dp++;
1657 				buflen -= sizeof(struct kinfo_proc);
1658 			}
1659 			needed += sizeof(struct kinfo_proc);
1660 		} else { /* KERN_PROC2 */
1661 			if (buflen >= elem_size && elem_count > 0) {
1662 				fill_kproc2(p, &kproc2);
1663 				/*
1664 				 * Copy out elem_size, but not larger than
1665 				 * the size of a struct kinfo_proc2.
1666 				 */
1667 				error = copyout(&kproc2, dp2,
1668 				    min(sizeof(kproc2), elem_size));
1669 				if (error)
1670 					goto cleanup;
1671 				dp2 += elem_size;
1672 				buflen -= elem_size;
1673 				elem_count--;
1674 			}
1675 			needed += elem_size;
1676 		}
1677 	}
1678 	pd++;
1679 	if (pd->pd_list != NULL)
1680 		goto again;
1681 	proclist_unlock_read();
1682 
1683 	if (where != NULL) {
1684 		if (type == KERN_PROC)
1685 			*sizep = (caddr_t)dp - where;
1686 		else
1687 			*sizep = dp2 - where;
1688 		if (needed > *sizep)
1689 			return (ENOMEM);
1690 	} else {
1691 		needed += KERN_PROCSLOP;
1692 		*sizep = needed;
1693 	}
1694 	return (0);
1695  cleanup:
1696 	proclist_unlock_read();
1697 	return (error);
1698 }
1699 
1700 
1701 /*
1702  * try over estimating by 5 LWPs
1703  */
1704 #define KERN_LWPSLOP	(5 * sizeof(struct kinfo_lwp))
1705 
1706 static int
1707 sysctl_dolwp(int *name, u_int namelen, void *vwhere, size_t *sizep)
1708 {
1709 	struct kinfo_lwp klwp;
1710 	struct proc *p;
1711 	struct lwp *l;
1712 	char *where, *dp;
1713 	int type, pid, elem_size, elem_count;
1714 	int buflen, needed, error;
1715 
1716 	dp = where = vwhere;
1717 	buflen = where != NULL ? *sizep : 0;
1718 	error = needed = 0;
1719 	type = name[0];
1720 
1721 	if (namelen != 4)
1722 		return (EINVAL);
1723 	pid = name[1];
1724 	elem_size = name[2];
1725 	elem_count = name[3];
1726 
1727 	p = pfind(pid);
1728 	if (p == NULL)
1729 		return (ESRCH);
1730 	LIST_FOREACH(l, &p->p_lwps, l_sibling) {
1731 		if (buflen >= elem_size && elem_count > 0) {
1732 			fill_lwp(l, &klwp);
1733 			/*
1734 			 * Copy out elem_size, but not larger than
1735 			 * the size of a struct kinfo_proc2.
1736 			 */
1737 			error = copyout(&klwp, dp,
1738 			    min(sizeof(klwp), elem_size));
1739 			if (error)
1740 				goto cleanup;
1741 			dp += elem_size;
1742 			buflen -= elem_size;
1743 			elem_count--;
1744 		}
1745 		needed += elem_size;
1746 	}
1747 
1748 	if (where != NULL) {
1749 		*sizep = dp - where;
1750 		if (needed > *sizep)
1751 			return (ENOMEM);
1752 	} else {
1753 		needed += KERN_PROCSLOP;
1754 		*sizep = needed;
1755 	}
1756 	return (0);
1757  cleanup:
1758 	return (error);
1759 }
1760 
1761 /*
1762  * Fill in an eproc structure for the specified process.
1763  */
1764 void
1765 fill_eproc(struct proc *p, struct eproc *ep)
1766 {
1767 	struct tty *tp;
1768 	struct lwp *l;
1769 
1770 	ep->e_paddr = p;
1771 	ep->e_sess = p->p_session;
1772 	ep->e_pcred = *p->p_cred;
1773 	ep->e_ucred = *p->p_ucred;
1774 	if (p->p_stat == SIDL || P_ZOMBIE(p)) {
1775 		ep->e_vm.vm_rssize = 0;
1776 		ep->e_vm.vm_tsize = 0;
1777 		ep->e_vm.vm_dsize = 0;
1778 		ep->e_vm.vm_ssize = 0;
1779 		/* ep->e_vm.vm_pmap = XXX; */
1780 	} else {
1781 		struct vmspace *vm = p->p_vmspace;
1782 
1783 		ep->e_vm.vm_rssize = vm_resident_count(vm);
1784 		ep->e_vm.vm_tsize = vm->vm_tsize;
1785 		ep->e_vm.vm_dsize = vm->vm_dsize;
1786 		ep->e_vm.vm_ssize = vm->vm_ssize;
1787 
1788 		/* Pick a "representative" LWP */
1789 		l = proc_representative_lwp(p);
1790 
1791 		if (l->l_wmesg)
1792 			strncpy(ep->e_wmesg, l->l_wmesg, WMESGLEN);
1793 	}
1794 	if (p->p_pptr)
1795 		ep->e_ppid = p->p_pptr->p_pid;
1796 	else
1797 		ep->e_ppid = 0;
1798 	ep->e_pgid = p->p_pgrp->pg_id;
1799 	ep->e_sid = ep->e_sess->s_sid;
1800 	ep->e_jobc = p->p_pgrp->pg_jobc;
1801 	if ((p->p_flag & P_CONTROLT) &&
1802 	    (tp = ep->e_sess->s_ttyp)) {
1803 		ep->e_tdev = tp->t_dev;
1804 		ep->e_tpgid = tp->t_pgrp ? tp->t_pgrp->pg_id : NO_PGID;
1805 		ep->e_tsess = tp->t_session;
1806 	} else
1807 		ep->e_tdev = NODEV;
1808 
1809 	ep->e_xsize = ep->e_xrssize = 0;
1810 	ep->e_xccount = ep->e_xswrss = 0;
1811 	ep->e_flag = ep->e_sess->s_ttyvp ? EPROC_CTTY : 0;
1812 	if (SESS_LEADER(p))
1813 		ep->e_flag |= EPROC_SLEADER;
1814 	strncpy(ep->e_login, ep->e_sess->s_login, MAXLOGNAME);
1815 }
1816 
1817 /*
1818  * Fill in an eproc structure for the specified process.
1819  */
1820 static void
1821 fill_kproc2(struct proc *p, struct kinfo_proc2 *ki)
1822 {
1823 	struct tty *tp;
1824 	struct lwp *l;
1825 	struct timeval ut, st;
1826 
1827 	memset(ki, 0, sizeof(*ki));
1828 
1829 	ki->p_paddr = PTRTOINT64(p);
1830 	ki->p_fd = PTRTOINT64(p->p_fd);
1831 	ki->p_cwdi = PTRTOINT64(p->p_cwdi);
1832 	ki->p_stats = PTRTOINT64(p->p_stats);
1833 	ki->p_limit = PTRTOINT64(p->p_limit);
1834 	ki->p_vmspace = PTRTOINT64(p->p_vmspace);
1835 	ki->p_sigacts = PTRTOINT64(p->p_sigacts);
1836 	ki->p_sess = PTRTOINT64(p->p_session);
1837 	ki->p_tsess = 0;	/* may be changed if controlling tty below */
1838 	ki->p_ru = PTRTOINT64(p->p_ru);
1839 
1840 	ki->p_eflag = 0;
1841 	ki->p_exitsig = p->p_exitsig;
1842 	ki->p_flag = p->p_flag;
1843 
1844 	ki->p_pid = p->p_pid;
1845 	if (p->p_pptr)
1846 		ki->p_ppid = p->p_pptr->p_pid;
1847 	else
1848 		ki->p_ppid = 0;
1849 	ki->p_sid = p->p_session->s_sid;
1850 	ki->p__pgid = p->p_pgrp->pg_id;
1851 
1852 	ki->p_tpgid = NO_PGID;	/* may be changed if controlling tty below */
1853 
1854 	ki->p_uid = p->p_ucred->cr_uid;
1855 	ki->p_ruid = p->p_cred->p_ruid;
1856 	ki->p_gid = p->p_ucred->cr_gid;
1857 	ki->p_rgid = p->p_cred->p_rgid;
1858 	ki->p_svuid = p->p_cred->p_svuid;
1859 	ki->p_svgid = p->p_cred->p_svgid;
1860 
1861 	memcpy(ki->p_groups, p->p_cred->pc_ucred->cr_groups,
1862 	    min(sizeof(ki->p_groups), sizeof(p->p_cred->pc_ucred->cr_groups)));
1863 	ki->p_ngroups = p->p_cred->pc_ucred->cr_ngroups;
1864 
1865 	ki->p_jobc = p->p_pgrp->pg_jobc;
1866 	if ((p->p_flag & P_CONTROLT) && (tp = p->p_session->s_ttyp)) {
1867 		ki->p_tdev = tp->t_dev;
1868 		ki->p_tpgid = tp->t_pgrp ? tp->t_pgrp->pg_id : NO_PGID;
1869 		ki->p_tsess = PTRTOINT64(tp->t_session);
1870 	} else {
1871 		ki->p_tdev = NODEV;
1872 	}
1873 
1874 	ki->p_estcpu = p->p_estcpu;
1875 	ki->p_rtime_sec = p->p_rtime.tv_sec;
1876 	ki->p_rtime_usec = p->p_rtime.tv_usec;
1877 	ki->p_cpticks = p->p_cpticks;
1878 	ki->p_pctcpu = p->p_pctcpu;
1879 
1880 	ki->p_uticks = p->p_uticks;
1881 	ki->p_sticks = p->p_sticks;
1882 	ki->p_iticks = p->p_iticks;
1883 
1884 	ki->p_tracep = PTRTOINT64(p->p_tracep);
1885 	ki->p_traceflag = p->p_traceflag;
1886 
1887 
1888 	memcpy(&ki->p_siglist, &p->p_sigctx.ps_siglist, sizeof(ki_sigset_t));
1889 	memcpy(&ki->p_sigmask, &p->p_sigctx.ps_sigmask, sizeof(ki_sigset_t));
1890 	memcpy(&ki->p_sigignore, &p->p_sigctx.ps_sigignore,sizeof(ki_sigset_t));
1891 	memcpy(&ki->p_sigcatch, &p->p_sigctx.ps_sigcatch, sizeof(ki_sigset_t));
1892 
1893 	ki->p_stat = p->p_stat; /* Will likely be overridden by LWP status */
1894 	ki->p_realstat = p->p_stat;
1895 	ki->p_nice = p->p_nice;
1896 
1897 	ki->p_xstat = p->p_xstat;
1898 	ki->p_acflag = p->p_acflag;
1899 
1900 	strncpy(ki->p_comm, p->p_comm,
1901 	    min(sizeof(ki->p_comm), sizeof(p->p_comm)));
1902 
1903 	strncpy(ki->p_login, p->p_session->s_login,
1904 	    min(sizeof ki->p_login - 1, sizeof p->p_session->s_login));
1905 
1906 	ki->p_nlwps = p->p_nlwps;
1907 	ki->p_nrlwps = p->p_nrlwps;
1908 	ki->p_realflag = p->p_flag;
1909 
1910 	if (p->p_stat == SIDL || P_ZOMBIE(p)) {
1911 		ki->p_vm_rssize = 0;
1912 		ki->p_vm_tsize = 0;
1913 		ki->p_vm_dsize = 0;
1914 		ki->p_vm_ssize = 0;
1915 		l = NULL;
1916 	} else {
1917 		struct vmspace *vm = p->p_vmspace;
1918 
1919 		ki->p_vm_rssize = vm_resident_count(vm);
1920 		ki->p_vm_tsize = vm->vm_tsize;
1921 		ki->p_vm_dsize = vm->vm_dsize;
1922 		ki->p_vm_ssize = vm->vm_ssize;
1923 
1924 		/* Pick a "representative" LWP */
1925 		l = proc_representative_lwp(p);
1926 		ki->p_forw = PTRTOINT64(l->l_forw);
1927 		ki->p_back = PTRTOINT64(l->l_back);
1928 		ki->p_addr = PTRTOINT64(l->l_addr);
1929 		ki->p_stat = l->l_stat;
1930 		ki->p_flag |= l->l_flag;
1931 		ki->p_swtime = l->l_swtime;
1932 		ki->p_slptime = l->l_slptime;
1933 		if (l->l_stat == LSONPROC) {
1934 			KDASSERT(l->l_cpu != NULL);
1935 			ki->p_schedflags = l->l_cpu->ci_schedstate.spc_flags;
1936 		} else
1937 			ki->p_schedflags = 0;
1938 		ki->p_holdcnt = l->l_holdcnt;
1939 		ki->p_priority = l->l_priority;
1940 		ki->p_usrpri = l->l_usrpri;
1941 		if (l->l_wmesg)
1942 			strncpy(ki->p_wmesg, l->l_wmesg, sizeof(ki->p_wmesg));
1943 		ki->p_wchan = PTRTOINT64(l->l_wchan);
1944 
1945 	}
1946 
1947 	if (p->p_session->s_ttyvp)
1948 		ki->p_eflag |= EPROC_CTTY;
1949 	if (SESS_LEADER(p))
1950 		ki->p_eflag |= EPROC_SLEADER;
1951 
1952 	/* XXX Is this double check necessary? */
1953 	if (P_ZOMBIE(p)) {
1954 		ki->p_uvalid = 0;
1955 	} else {
1956 		ki->p_uvalid = 1;
1957 
1958 		ki->p_ustart_sec = p->p_stats->p_start.tv_sec;
1959 		ki->p_ustart_usec = p->p_stats->p_start.tv_usec;
1960 
1961 		calcru(p, &ut, &st, 0);
1962 		ki->p_uutime_sec = ut.tv_sec;
1963 		ki->p_uutime_usec = ut.tv_usec;
1964 		ki->p_ustime_sec = st.tv_sec;
1965 		ki->p_ustime_usec = st.tv_usec;
1966 
1967 		ki->p_uru_maxrss = p->p_stats->p_ru.ru_maxrss;
1968 		ki->p_uru_ixrss = p->p_stats->p_ru.ru_ixrss;
1969 		ki->p_uru_idrss = p->p_stats->p_ru.ru_idrss;
1970 		ki->p_uru_isrss = p->p_stats->p_ru.ru_isrss;
1971 		ki->p_uru_minflt = p->p_stats->p_ru.ru_minflt;
1972 		ki->p_uru_majflt = p->p_stats->p_ru.ru_majflt;
1973 		ki->p_uru_nswap = p->p_stats->p_ru.ru_nswap;
1974 		ki->p_uru_inblock = p->p_stats->p_ru.ru_inblock;
1975 		ki->p_uru_oublock = p->p_stats->p_ru.ru_oublock;
1976 		ki->p_uru_msgsnd = p->p_stats->p_ru.ru_msgsnd;
1977 		ki->p_uru_msgrcv = p->p_stats->p_ru.ru_msgrcv;
1978 		ki->p_uru_nsignals = p->p_stats->p_ru.ru_nsignals;
1979 		ki->p_uru_nvcsw = p->p_stats->p_ru.ru_nvcsw;
1980 		ki->p_uru_nivcsw = p->p_stats->p_ru.ru_nivcsw;
1981 
1982 		timeradd(&p->p_stats->p_cru.ru_utime,
1983 			 &p->p_stats->p_cru.ru_stime, &ut);
1984 		ki->p_uctime_sec = ut.tv_sec;
1985 		ki->p_uctime_usec = ut.tv_usec;
1986 	}
1987 #ifdef MULTIPROCESSOR
1988 	if (l && l->l_cpu != NULL)
1989 		ki->p_cpuid = l->l_cpu->ci_cpuid;
1990 	else
1991 #endif
1992 		ki->p_cpuid = KI_NOCPU;
1993 
1994 }
1995 
1996 /*
1997  * Fill in a kinfo_lwp structure for the specified lwp.
1998  */
1999 static void
2000 fill_lwp(struct lwp *l, struct kinfo_lwp *kl)
2001 {
2002 
2003 	kl->l_forw = PTRTOINT64(l->l_forw);
2004 	kl->l_back = PTRTOINT64(l->l_back);
2005 	kl->l_laddr = PTRTOINT64(l);
2006 	kl->l_addr = PTRTOINT64(l->l_addr);
2007 	kl->l_stat = l->l_stat;
2008 	kl->l_lid = l->l_lid;
2009 	kl->l_flag = l->l_flag;
2010 
2011 	kl->l_swtime = l->l_swtime;
2012 	kl->l_slptime = l->l_slptime;
2013 	if (l->l_stat == LSONPROC) {
2014 		KDASSERT(l->l_cpu != NULL);
2015 		kl->l_schedflags = l->l_cpu->ci_schedstate.spc_flags;
2016 	} else
2017 		kl->l_schedflags = 0;
2018 	kl->l_holdcnt = l->l_holdcnt;
2019 	kl->l_priority = l->l_priority;
2020 	kl->l_usrpri = l->l_usrpri;
2021 	if (l->l_wmesg)
2022 		strncpy(kl->l_wmesg, l->l_wmesg, sizeof(kl->l_wmesg));
2023 	kl->l_wchan = PTRTOINT64(l->l_wchan);
2024 #ifdef MULTIPROCESSOR
2025 	if (l->l_cpu != NULL)
2026 		kl->l_cpuid = l->l_cpu->ci_cpuid;
2027 	else
2028 #endif
2029 		kl->l_cpuid = KI_NOCPU;
2030 }
2031 
2032 int
2033 sysctl_procargs(int *name, u_int namelen, void *where, size_t *sizep,
2034     struct proc *up)
2035 {
2036 	struct ps_strings pss;
2037 	struct proc *p;
2038 	size_t len, upper_bound, xlen, i;
2039 	struct uio auio;
2040 	struct iovec aiov;
2041 	vaddr_t argv;
2042 	pid_t pid;
2043 	int nargv, type, error;
2044 	char *arg;
2045 	char *tmp;
2046 
2047 	if (namelen != 2)
2048 		return (EINVAL);
2049 	pid = name[0];
2050 	type = name[1];
2051 
2052 	switch (type) {
2053 	case KERN_PROC_ARGV:
2054 	case KERN_PROC_NARGV:
2055 	case KERN_PROC_ENV:
2056 	case KERN_PROC_NENV:
2057 		/* ok */
2058 		break;
2059 	default:
2060 		return (EINVAL);
2061 	}
2062 
2063 	/* check pid */
2064 	if ((p = pfind(pid)) == NULL)
2065 		return (EINVAL);
2066 
2067 	/* only root or same user change look at the environment */
2068 	if (type == KERN_PROC_ENV || type == KERN_PROC_NENV) {
2069 		if (up->p_ucred->cr_uid != 0) {
2070 			if (up->p_cred->p_ruid != p->p_cred->p_ruid ||
2071 			    up->p_cred->p_ruid != p->p_cred->p_svuid)
2072 				return (EPERM);
2073 		}
2074 	}
2075 
2076 	if (sizep != NULL && where == NULL) {
2077 		if (type == KERN_PROC_NARGV || type == KERN_PROC_NENV)
2078 			*sizep = sizeof (int);
2079 		else
2080 			*sizep = ARG_MAX;	/* XXX XXX XXX */
2081 		return (0);
2082 	}
2083 	if (where == NULL || sizep == NULL)
2084 		return (EINVAL);
2085 
2086 	/*
2087 	 * Zombies don't have a stack, so we can't read their psstrings.
2088 	 * System processes also don't have a user stack.
2089 	 */
2090 	if (P_ZOMBIE(p) || (p->p_flag & P_SYSTEM) != 0)
2091 		return (EINVAL);
2092 
2093 	/*
2094 	 * Lock the process down in memory.
2095 	 */
2096 	/* XXXCDC: how should locking work here? */
2097 	if ((p->p_flag & P_WEXIT) || (p->p_vmspace->vm_refcnt < 1))
2098 		return (EFAULT);
2099 
2100 	p->p_vmspace->vm_refcnt++;	/* XXX */
2101 
2102 	/*
2103 	 * Allocate a temporary buffer to hold the arguments.
2104 	 */
2105 	arg = malloc(PAGE_SIZE, M_TEMP, M_WAITOK);
2106 
2107 	/*
2108 	 * Read in the ps_strings structure.
2109 	 */
2110 	aiov.iov_base = &pss;
2111 	aiov.iov_len = sizeof(pss);
2112 	auio.uio_iov = &aiov;
2113 	auio.uio_iovcnt = 1;
2114 	auio.uio_offset = (vaddr_t)p->p_psstr;
2115 	auio.uio_resid = sizeof(pss);
2116 	auio.uio_segflg = UIO_SYSSPACE;
2117 	auio.uio_rw = UIO_READ;
2118 	auio.uio_procp = NULL;
2119 	error = uvm_io(&p->p_vmspace->vm_map, &auio);
2120 	if (error)
2121 		goto done;
2122 
2123 	if (type == KERN_PROC_ARGV || type == KERN_PROC_NARGV)
2124 		memcpy(&nargv, (char *)&pss + p->p_psnargv, sizeof(nargv));
2125 	else
2126 		memcpy(&nargv, (char *)&pss + p->p_psnenv, sizeof(nargv));
2127 	if (type == KERN_PROC_NARGV || type == KERN_PROC_NENV) {
2128 		error = copyout(&nargv, where, sizeof(nargv));
2129 		*sizep = sizeof(nargv);
2130 		goto done;
2131 	}
2132 	/*
2133 	 * Now read the address of the argument vector.
2134 	 */
2135 	switch (type) {
2136 	case KERN_PROC_ARGV:
2137 		/* XXX compat32 stuff here */
2138 		memcpy(&tmp, (char *)&pss + p->p_psargv, sizeof(tmp));
2139 		break;
2140 	case KERN_PROC_ENV:
2141 		memcpy(&tmp, (char *)&pss + p->p_psenv, sizeof(tmp));
2142 		break;
2143 	default:
2144 		return (EINVAL);
2145 	}
2146 	auio.uio_offset = (off_t)(long)tmp;
2147 	aiov.iov_base = &argv;
2148 	aiov.iov_len = sizeof(argv);
2149 	auio.uio_iov = &aiov;
2150 	auio.uio_iovcnt = 1;
2151 	auio.uio_resid = sizeof(argv);
2152 	auio.uio_segflg = UIO_SYSSPACE;
2153 	auio.uio_rw = UIO_READ;
2154 	auio.uio_procp = NULL;
2155 	error = uvm_io(&p->p_vmspace->vm_map, &auio);
2156 	if (error)
2157 		goto done;
2158 
2159 	/*
2160 	 * Now copy in the actual argument vector, one page at a time,
2161 	 * since we don't know how long the vector is (though, we do
2162 	 * know how many NUL-terminated strings are in the vector).
2163 	 */
2164 	len = 0;
2165 	upper_bound = *sizep;
2166 	for (; nargv != 0 && len < upper_bound; len += xlen) {
2167 		aiov.iov_base = arg;
2168 		aiov.iov_len = PAGE_SIZE;
2169 		auio.uio_iov = &aiov;
2170 		auio.uio_iovcnt = 1;
2171 		auio.uio_offset = argv + len;
2172 		xlen = PAGE_SIZE - ((argv + len) & PAGE_MASK);
2173 		auio.uio_resid = xlen;
2174 		auio.uio_segflg = UIO_SYSSPACE;
2175 		auio.uio_rw = UIO_READ;
2176 		auio.uio_procp = NULL;
2177 		error = uvm_io(&p->p_vmspace->vm_map, &auio);
2178 		if (error)
2179 			goto done;
2180 
2181 		for (i = 0; i < xlen && nargv != 0; i++) {
2182 			if (arg[i] == '\0')
2183 				nargv--;	/* one full string */
2184 		}
2185 
2186 		/*
2187 		 * Make sure we don't copyout past the end of the user's
2188 		 * buffer.
2189 		 */
2190 		if (len + i > upper_bound)
2191 			i = upper_bound - len;
2192 
2193 		error = copyout(arg, (char *)where + len, i);
2194 		if (error)
2195 			break;
2196 
2197 		if (nargv == 0) {
2198 			len += i;
2199 			break;
2200 		}
2201 	}
2202 	*sizep = len;
2203 
2204 done:
2205 	uvmspace_free(p->p_vmspace);
2206 
2207 	free(arg, M_TEMP);
2208 	return (error);
2209 }
2210 
2211 #if NPTY > 0
2212 int pty_maxptys(int, int);		/* defined in kern/tty_pty.c */
2213 
2214 /*
2215  * Validate parameters and get old / set new parameters
2216  * for pty sysctl function.
2217  */
2218 static int
2219 sysctl_pty(void *oldp, size_t *oldlenp, void *newp, size_t newlen)
2220 {
2221 	int error = 0;
2222 	int oldmax = 0, newmax = 0;
2223 
2224 	/* get current value of maxptys */
2225 	oldmax = pty_maxptys(0, 0);
2226 
2227 	SYSCTL_SCALAR_CORE_TYP(oldp, oldlenp, &oldmax, int)
2228 
2229 	if (!error && newp) {
2230 		SYSCTL_SCALAR_NEWPCHECK_TYP(newp, newlen, int)
2231 		SYSCTL_SCALAR_NEWPCOP_TYP(newp, &newmax, int)
2232 
2233 		if (newmax != pty_maxptys(newmax, (newp != NULL)))
2234 			return (EINVAL);
2235 
2236 	}
2237 
2238 	return (error);
2239 }
2240 #endif /* NPTY > 0 */
2241 
2242 static int
2243 sysctl_dotkstat(int *name, u_int namelen, void *where, size_t *sizep,
2244     void *newp)
2245 {
2246 
2247 	/* all sysctl names at this level are terminal */
2248 	if (namelen != 1)
2249 		return (ENOTDIR);		/* overloaded */
2250 
2251 	switch (name[0]) {
2252 	case KERN_TKSTAT_NIN:
2253 		return (sysctl_rdquad(where, sizep, newp, tk_nin));
2254 	case KERN_TKSTAT_NOUT:
2255 		return (sysctl_rdquad(where, sizep, newp, tk_nout));
2256 	case KERN_TKSTAT_CANCC:
2257 		return (sysctl_rdquad(where, sizep, newp, tk_cancc));
2258 	case KERN_TKSTAT_RAWCC:
2259 		return (sysctl_rdquad(where, sizep, newp, tk_rawcc));
2260 	default:
2261 		return (EOPNOTSUPP);
2262 	}
2263 }
2264