xref: /netbsd-src/sys/kern/kern_sysctl.c (revision 001c68bd94f75ce9270b69227c4199fbf34ee396)
1 /*	$NetBSD: kern_sysctl.c,v 1.138 2003/07/02 20:07:45 ragge 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 1.138 2003/07/02 20:07:45 ragge 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_SOMAXKVA:
590 	    {
591 		int new_somaxkva = somaxkva;
592 
593 		error = sysctl_int(oldp, oldlenp, newp, newlen, &new_somaxkva);
594 		if (newp && !error) {
595 			if (new_somaxkva < (16 * 1024 * 1024)) /* sanity */
596 				return (EINVAL);
597 			somaxkva = new_somaxkva;
598 		}
599 		return (error);
600 	    }
601 	case KERN_SBMAX:
602 	    {
603 		int new_sbmax = sb_max;
604 
605 		error = sysctl_int(oldp, oldlenp, newp, newlen, &new_sbmax);
606 		if (newp && !error) {
607 			if (new_sbmax < (16 * 1024)) /* sanity */
608 				return (EINVAL);
609 			sb_max = new_sbmax;
610 		}
611 		return (error);
612 	    }
613 	case KERN_TKSTAT:
614 		return (sysctl_dotkstat(name + 1, namelen - 1, oldp, oldlenp,
615 		    newp));
616 	case KERN_MONOTONIC_CLOCK:	/* XXX _POSIX_VERSION */
617 		return (sysctl_rdint(oldp, oldlenp, newp, 200112));
618 	case KERN_URND:
619 #if NRND > 0
620 		if (rnd_extract_data(&v, sizeof(v), RND_EXTRACT_ANY) ==
621 		    sizeof(v))
622 			return (sysctl_rdint(oldp, oldlenp, newp, v));
623 		else
624 			return (EIO);	/*XXX*/
625 #else
626 		return (EOPNOTSUPP);
627 #endif
628 	case KERN_LABELSECTOR:
629 		return (sysctl_rdint(oldp, oldlenp, newp, LABELSECTOR));
630 	case KERN_LABELOFFSET:
631 		return (sysctl_rdint(oldp, oldlenp, newp, LABELOFFSET));
632 	case KERN_FORKFSLEEP:
633 	    {
634 		/* userland sees value in ms, internally is in ticks */
635 		int timo, lsleep = forkfsleep * 1000 / hz;
636 
637 		error = sysctl_int(oldp, oldlenp, newp, newlen, &lsleep);
638 		if (newp && !error) {
639 			/* refuse negative values, and overly 'long time' */
640 			if (lsleep < 0 || lsleep > MAXSLP * 1000)
641 				return (EINVAL);
642 
643 			timo = mstohz(lsleep);
644 
645 			/* if the interval is >0 ms && <1 tick, use 1 tick */
646 			if (lsleep != 0 && timo == 0)
647 				forkfsleep = 1;
648 			else
649 				forkfsleep = timo;
650 		}
651 		return (error);
652 	    }
653 	case KERN_POSIX_THREADS:	/* XXX _POSIX_VERSION */
654 		return (sysctl_rdint(oldp, oldlenp, newp, 200112));
655 	case KERN_POSIX_SEMAPHORES:	/* XXX _POSIX_VERSION */
656 #ifdef P1003_1B_SEMAPHORE
657 		return (sysctl_rdint(oldp, oldlenp, newp, 200112));
658 #else
659 		return (sysctl_rdint(oldp, oldlenp, newp, 0));
660 #endif
661 	case KERN_POSIX_BARRIERS:	/* XXX _POSIX_VERSION */
662 		return (sysctl_rdint(oldp, oldlenp, newp, 200112));
663 	case KERN_POSIX_TIMERS:		/* XXX _POSIX_VERSION */
664 		return (sysctl_rdint(oldp, oldlenp, newp, 200112));
665 	case KERN_POSIX_SPIN_LOCKS:	/* XXX _POSIX_VERSION */
666 		return (sysctl_rdint(oldp, oldlenp, newp, 200112));
667 	case KERN_POSIX_READER_WRITER_LOCKS:	/* XXX _POSIX_VERSION */
668 		return (sysctl_rdint(oldp, oldlenp, newp, 200112));
669 	case KERN_DUMP_ON_PANIC:
670 		return (sysctl_int(oldp, oldlenp, newp, newlen, &dumponpanic));
671 
672 	default:
673 		return (EOPNOTSUPP);
674 	}
675 	/* NOTREACHED */
676 }
677 
678 /*
679  * hardware related system variables.
680  */
681 int
682 hw_sysctl(int *name, u_int namelen, void *oldp, size_t *oldlenp,
683     void *newp, size_t newlen, struct proc *p)
684 {
685 
686 	/* All sysctl names at this level, except for a few, are terminal. */
687 	switch (name[0]) {
688 	case HW_DISKSTATS:
689 		/* Not terminal. */
690 		break;
691 	default:
692 		if (namelen != 1)
693 			return (ENOTDIR);	/* overloaded */
694 	}
695 
696 	switch (name[0]) {
697 	case HW_MACHINE:
698 		return (sysctl_rdstring(oldp, oldlenp, newp, machine));
699 	case HW_MACHINE_ARCH:
700 		return (sysctl_rdstring(oldp, oldlenp, newp, machine_arch));
701 	case HW_MODEL:
702 		return (sysctl_rdstring(oldp, oldlenp, newp, cpu_model));
703 	case HW_NCPU:
704 		return (sysctl_rdint(oldp, oldlenp, newp, sysctl_ncpus()));
705 	case HW_BYTEORDER:
706 		return (sysctl_rdint(oldp, oldlenp, newp, BYTE_ORDER));
707 	case HW_PHYSMEM:
708 	    {
709 		u_int rval;
710 
711 		if ((u_int)physmem > (UINT_MAX / PAGE_SIZE))
712 			rval = UINT_MAX;
713 		else
714 			rval = physmem * PAGE_SIZE;
715 		return (sysctl_rdint(oldp, oldlenp, newp, rval));
716 	    }
717 	case HW_PHYSMEM64:
718 		return (sysctl_rdquad(oldp, oldlenp, newp,
719 		    (u_quad_t)physmem * PAGE_SIZE));
720 	case HW_USERMEM:
721 	    {
722 		u_int rval;
723 
724 		if ((u_int)(physmem - uvmexp.wired) > (UINT_MAX / PAGE_SIZE))
725 			rval = UINT_MAX;
726 		else
727 			rval = (physmem - uvmexp.wired) * PAGE_SIZE;
728 		return (sysctl_rdint(oldp, oldlenp, newp, rval));
729 	    }
730 	case HW_USERMEM64:
731 		return (sysctl_rdquad(oldp, oldlenp, newp,
732 		    (u_quad_t)(physmem - uvmexp.wired) * PAGE_SIZE));
733 	case HW_PAGESIZE:
734 		return (sysctl_rdint(oldp, oldlenp, newp, PAGE_SIZE));
735 	case HW_ALIGNBYTES:
736 		return (sysctl_rdint(oldp, oldlenp, newp, ALIGNBYTES));
737 	case HW_DISKNAMES:
738 		return (sysctl_disknames(oldp, oldlenp));
739 	case HW_DISKSTATS:
740 		return (sysctl_diskstats(name + 1, namelen - 1, oldp, oldlenp));
741 	case HW_CNMAGIC: {
742 		char magic[CNS_LEN];
743 		int error;
744 
745 		if (oldp)
746 			cn_get_magic(magic, CNS_LEN);
747 		error = sysctl_string(oldp, oldlenp, newp, newlen,
748 		    magic, sizeof(magic));
749 		if (newp && !error) {
750 			error = cn_set_magic(magic);
751 		}
752 		return (error);
753 	}
754 	default:
755 		return (EOPNOTSUPP);
756 	}
757 	/* NOTREACHED */
758 }
759 
760 #ifdef DEBUG
761 /*
762  * Debugging related system variables.
763  */
764 struct ctldebug /* debug0, */ /* debug1, */ debug2, debug3, debug4;
765 struct ctldebug debug5, debug6, debug7, debug8, debug9;
766 struct ctldebug debug10, debug11, debug12, debug13, debug14;
767 struct ctldebug debug15, debug16, debug17, debug18, debug19;
768 static struct ctldebug *debugvars[CTL_DEBUG_MAXID] = {
769 	&debug0, &debug1, &debug2, &debug3, &debug4,
770 	&debug5, &debug6, &debug7, &debug8, &debug9,
771 	&debug10, &debug11, &debug12, &debug13, &debug14,
772 	&debug15, &debug16, &debug17, &debug18, &debug19,
773 };
774 
775 int
776 debug_sysctl(int *name, u_int namelen, void *oldp, size_t *oldlenp,
777     void *newp, size_t newlen, struct proc *p)
778 {
779 	struct ctldebug *cdp;
780 
781 	/* all sysctl names at this level are name and field */
782 	if (namelen != 2)
783 		return (ENOTDIR);		/* overloaded */
784 	if (name[0] >= CTL_DEBUG_MAXID)
785 		return (EOPNOTSUPP);
786 	cdp = debugvars[name[0]];
787 	if (cdp->debugname == 0)
788 		return (EOPNOTSUPP);
789 	switch (name[1]) {
790 	case CTL_DEBUG_NAME:
791 		return (sysctl_rdstring(oldp, oldlenp, newp, cdp->debugname));
792 	case CTL_DEBUG_VALUE:
793 		return (sysctl_int(oldp, oldlenp, newp, newlen, cdp->debugvar));
794 	default:
795 		return (EOPNOTSUPP);
796 	}
797 	/* NOTREACHED */
798 }
799 #endif /* DEBUG */
800 
801 int
802 proc_sysctl(int *name, u_int namelen, void *oldp, size_t *oldlenp,
803     void *newp, size_t newlen, struct proc *p)
804 {
805 	struct proc *ptmp = NULL;
806 	const struct proclist_desc *pd;
807 	int error = 0;
808 	struct rlimit alim;
809 	struct plimit *newplim;
810 	char *tmps = NULL;
811 	size_t len, curlen;
812 	u_int i;
813 
814 	if (namelen < 2)
815 		return (EINVAL);
816 
817 	if (name[0] == PROC_CURPROC) {
818 		ptmp = p;
819 	} else {
820 		proclist_lock_read();
821 		for (pd = proclists; pd->pd_list != NULL; pd++) {
822 			for (ptmp = LIST_FIRST(pd->pd_list); ptmp != NULL;
823 			    ptmp = LIST_NEXT(ptmp, p_list)) {
824 				/* Skip embryonic processes. */
825 				if (ptmp->p_stat == SIDL)
826 					continue;
827 				if (ptmp->p_pid == (pid_t)name[0])
828 					break;
829 			}
830 			if (ptmp != NULL)
831 				break;
832 		}
833 		proclist_unlock_read();
834 		if (ptmp == NULL)
835 			return (ESRCH);
836 		if (p->p_ucred->cr_uid != 0) {
837 			if (p->p_cred->p_ruid != ptmp->p_cred->p_ruid ||
838 			    p->p_cred->p_ruid != ptmp->p_cred->p_svuid)
839 				return (EPERM);
840 			if (ptmp->p_cred->p_rgid != ptmp->p_cred->p_svgid)
841 				return (EPERM); /* sgid proc */
842 			for (i = 0; i < p->p_ucred->cr_ngroups; i++) {
843 				if (p->p_ucred->cr_groups[i] ==
844 				    ptmp->p_cred->p_rgid)
845 					break;
846 			}
847 			if (i == p->p_ucred->cr_ngroups)
848 				return (EPERM);
849 		}
850 	}
851 	switch (name[1]) {
852 	case PROC_PID_STOPFORK:
853 		if (namelen != 2)
854 			return (EINVAL);
855 		i = ((ptmp->p_flag & P_STOPFORK) != 0);
856 		if ((error = sysctl_int(oldp, oldlenp, newp, newlen, &i)) != 0)
857 			return (error);
858 		if (i != 0)
859 			ptmp->p_flag |= P_STOPFORK;
860 		else
861 			ptmp->p_flag &= ~P_STOPFORK;
862 		return (0);
863 		break;
864 
865 	case PROC_PID_STOPEXEC:
866 		if (namelen != 2)
867 			return (EINVAL);
868 		i = ((ptmp->p_flag & P_STOPEXEC) != 0);
869 		if ((error = sysctl_int(oldp, oldlenp, newp, newlen, &i)) != 0)
870 			return (error);
871 		if (i != 0)
872 			ptmp->p_flag |= P_STOPEXEC;
873 		else
874 			ptmp->p_flag &= ~P_STOPEXEC;
875 		return (0);
876 		break;
877 
878 	case PROC_PID_CORENAME:
879 		if (namelen != 2)
880 			return (EINVAL);
881 		/*
882 		 * Can't use sysctl_string() here because we may malloc a new
883 		 * area during the process, so we have to do it by hand.
884 		 */
885 		curlen = strlen(ptmp->p_limit->pl_corename) + 1;
886 		if (oldlenp && *oldlenp < curlen) {
887 			if (!oldp)
888 				*oldlenp = curlen;
889 			return (ENOMEM);
890 		}
891 		if (newp) {
892 			if (securelevel > 2)
893 				return (EPERM);
894 			if (newlen > MAXPATHLEN)
895 				return (ENAMETOOLONG);
896 			tmps = malloc(newlen + 1, M_TEMP, M_WAITOK);
897 			if (tmps == NULL)
898 				return (ENOMEM);
899 			error = copyin(newp, tmps, newlen + 1);
900 			tmps[newlen] = '\0';
901 			if (error)
902 				goto cleanup;
903 			/* Enforce to be either 'core' for end with '.core' */
904 			if (newlen < 4) {	/* c.o.r.e */
905 				error = EINVAL;
906 				goto cleanup;
907 			}
908 			len = newlen - 4;
909 			if (len > 0) {
910 				if (tmps[len - 1] != '.' &&
911 				    tmps[len - 1] != '/') {
912 					error = EINVAL;
913 					goto cleanup;
914 				}
915 			}
916 			if (strcmp(&tmps[len], "core") != 0) {
917 				error = EINVAL;
918 				goto cleanup;
919 			}
920 		}
921 		if (oldp && oldlenp) {
922 			*oldlenp = curlen;
923 			error = copyout(ptmp->p_limit->pl_corename, oldp,
924 			    curlen);
925 		}
926 		if (newp && error == 0) {
927 			/* if the 2 strings are identical, don't limcopy() */
928 			if (strcmp(tmps, ptmp->p_limit->pl_corename) == 0) {
929 				error = 0;
930 				goto cleanup;
931 			}
932 			if (ptmp->p_limit->p_refcnt > 1 &&
933 			    (ptmp->p_limit->p_lflags & PL_SHAREMOD) == 0) {
934 				newplim = limcopy(ptmp->p_limit);
935 				limfree(ptmp->p_limit);
936 				ptmp->p_limit = newplim;
937 			}
938 			if (ptmp->p_limit->pl_corename != defcorename) {
939 				free(ptmp->p_limit->pl_corename, M_TEMP);
940 			}
941 			ptmp->p_limit->pl_corename = tmps;
942 			return (0);
943 		}
944 cleanup:
945 		if (tmps)
946 			free(tmps, M_TEMP);
947 		return (error);
948 		break;
949 
950 	case PROC_PID_LIMIT:
951 		if (namelen != 4 || name[2] >= PROC_PID_LIMIT_MAXID)
952 			return (EINVAL);
953 		memcpy(&alim, &ptmp->p_rlimit[name[2] - 1], sizeof(alim));
954 		if (name[3] == PROC_PID_LIMIT_TYPE_HARD)
955 			error = sysctl_quad(oldp, oldlenp, newp, newlen,
956 			    &alim.rlim_max);
957 		else if (name[3] == PROC_PID_LIMIT_TYPE_SOFT)
958 			error = sysctl_quad(oldp, oldlenp, newp, newlen,
959 			    &alim.rlim_cur);
960 		else
961 			error = (EINVAL);
962 
963 		if (error)
964 			return (error);
965 
966 		if (newp)
967 			error = dosetrlimit(ptmp, p->p_cred,
968 			    name[2] - 1, &alim);
969 		return (error);
970 		break;
971 
972 	default:
973 		return (EINVAL);
974 		break;
975 	}
976 	/* NOTREACHED */
977 	return (EINVAL);
978 }
979 
980 int
981 emul_sysctl(int *name, u_int namelen, void *oldp, size_t *oldlenp,
982     void *newp, size_t newlen, struct proc *p)
983 {
984 	static struct {
985 		const char *name;
986 		int type;
987 	} emulations[] = CTL_EMUL_NAMES;
988 	const struct emul *e;
989 	const char *ename;
990 #ifdef LKM
991 	extern struct lock exec_lock;	/* XXX */
992 	int error;
993 #else
994 	extern int nexecs_builtin;
995 	extern const struct execsw execsw_builtin[];
996 	int i;
997 #endif
998 
999 	/* all sysctl names at this level are name and field */
1000 	if (namelen < 2)
1001 		return (ENOTDIR);		/* overloaded */
1002 
1003 	if ((u_int) name[0] >= EMUL_MAXID || name[0] == 0)
1004 		return (EOPNOTSUPP);
1005 
1006 	ename = emulations[name[0]].name;
1007 
1008 #ifdef LKM
1009 	lockmgr(&exec_lock, LK_SHARED, NULL);
1010 	if ((e = emul_search(ename))) {
1011 		error = (*e->e_sysctl)(name + 1, namelen - 1, oldp, oldlenp,
1012 		    newp, newlen, p);
1013 	} else
1014 		error = EOPNOTSUPP;
1015 	lockmgr(&exec_lock, LK_RELEASE, NULL);
1016 
1017 	return (error);
1018 #else
1019 	for (i = 0; i < nexecs_builtin; i++) {
1020 		e = execsw_builtin[i].es_emul;
1021 		/*
1022 		 * In order to match e.g. e->e_name "irix o32"
1023 		 * with ename "irix", we limit the comparison
1024 		 * to the length of ename.
1025 		 */
1026 		if (e == NULL ||
1027 		    strncmp(ename, e->e_name, strlen(ename)) != 0 ||
1028 		    e->e_sysctl == NULL)
1029 			continue;
1030 
1031 		return ((*e->e_sysctl)(name + 1, namelen - 1, oldp, oldlenp,
1032 		    newp, newlen, p));
1033 	}
1034 
1035 	return (EOPNOTSUPP);
1036 #endif
1037 }
1038 /*
1039  * Convenience macros.
1040  */
1041 
1042 #define SYSCTL_SCALAR_CORE_LEN(oldp, oldlenp, valp, len) 		\
1043 	if (oldlenp) {							\
1044 		if (!oldp)						\
1045 			*oldlenp = len;					\
1046 		else {							\
1047 			if (*oldlenp < len)				\
1048 				return (ENOMEM);			\
1049 			*oldlenp = len;					\
1050 			error = copyout((caddr_t)valp, oldp, len);	\
1051 		}							\
1052 	}
1053 
1054 #define SYSCTL_SCALAR_CORE_TYP(oldp, oldlenp, valp, typ) \
1055 	SYSCTL_SCALAR_CORE_LEN(oldp, oldlenp, valp, sizeof(typ))
1056 
1057 #define SYSCTL_SCALAR_NEWPCHECK_LEN(newp, newlen, len)	\
1058 	if (newp && newlen != len)			\
1059 		return (EINVAL);
1060 
1061 #define SYSCTL_SCALAR_NEWPCHECK_TYP(newp, newlen, typ)	\
1062 	SYSCTL_SCALAR_NEWPCHECK_LEN(newp, newlen, sizeof(typ))
1063 
1064 #define SYSCTL_SCALAR_NEWPCOP_LEN(newp, valp, len)	\
1065 	if (error == 0 && newp)				\
1066 		error = copyin(newp, valp, len);
1067 
1068 #define SYSCTL_SCALAR_NEWPCOP_TYP(newp, valp, typ)	\
1069 	SYSCTL_SCALAR_NEWPCOP_LEN(newp, valp, sizeof(typ))
1070 
1071 #define SYSCTL_STRING_CORE(oldp, oldlenp, str)		\
1072 	if (oldlenp) {					\
1073 		len = strlen(str) + 1;			\
1074 		if (!oldp)				\
1075 			*oldlenp = len;			\
1076 		else {					\
1077 			if (*oldlenp < len) {		\
1078 				err2 = ENOMEM;		\
1079 				len = *oldlenp;		\
1080 			} else				\
1081 				*oldlenp = len;		\
1082 			error = copyout(str, oldp, len);\
1083 			if (error == 0)			\
1084 				error = err2;		\
1085 		}					\
1086 	}
1087 
1088 /*
1089  * Validate parameters and get old / set new parameters
1090  * for an integer-valued sysctl function.
1091  */
1092 int
1093 sysctl_int(void *oldp, size_t *oldlenp, void *newp, size_t newlen, int *valp)
1094 {
1095 	int error = 0;
1096 
1097 	SYSCTL_SCALAR_NEWPCHECK_TYP(newp, newlen, int)
1098 	SYSCTL_SCALAR_CORE_TYP(oldp, oldlenp, valp, int)
1099 	SYSCTL_SCALAR_NEWPCOP_TYP(newp, valp, int)
1100 
1101 	return (error);
1102 }
1103 
1104 
1105 /*
1106  * As above, but read-only.
1107  */
1108 int
1109 sysctl_rdint(void *oldp, size_t *oldlenp, void *newp, int val)
1110 {
1111 	int error = 0;
1112 
1113 	if (newp)
1114 		return (EPERM);
1115 
1116 	SYSCTL_SCALAR_CORE_TYP(oldp, oldlenp, &val, int)
1117 
1118 	return (error);
1119 }
1120 
1121 /*
1122  * Validate parameters and get old / set new parameters
1123  * for an quad-valued sysctl function.
1124  */
1125 int
1126 sysctl_quad(void *oldp, size_t *oldlenp, void *newp, size_t newlen,
1127     quad_t *valp)
1128 {
1129 	int error = 0;
1130 
1131 	SYSCTL_SCALAR_NEWPCHECK_TYP(newp, newlen, quad_t)
1132 	SYSCTL_SCALAR_CORE_TYP(oldp, oldlenp, valp, quad_t)
1133 	SYSCTL_SCALAR_NEWPCOP_TYP(newp, valp, quad_t)
1134 
1135 	return (error);
1136 }
1137 
1138 /*
1139  * As above, but read-only.
1140  */
1141 int
1142 sysctl_rdquad(void *oldp, size_t *oldlenp, void *newp, quad_t val)
1143 {
1144 	int error = 0;
1145 
1146 	if (newp)
1147 		return (EPERM);
1148 
1149 	SYSCTL_SCALAR_CORE_TYP(oldp, oldlenp, &val, quad_t)
1150 
1151 	return (error);
1152 }
1153 
1154 /*
1155  * Validate parameters and get old / set new parameters
1156  * for a string-valued sysctl function.
1157  */
1158 int
1159 sysctl_string(void *oldp, size_t *oldlenp, void *newp, size_t newlen, char *str,
1160     size_t maxlen)
1161 {
1162 	int error = 0, err2 = 0;
1163 	size_t len;
1164 
1165 	if (newp && newlen >= maxlen)
1166 		return (EINVAL);
1167 
1168 	SYSCTL_STRING_CORE(oldp, oldlenp, str);
1169 
1170 	if (error == 0 && newp) {
1171 		error = copyin(newp, str, newlen);
1172 		str[newlen] = 0;
1173 	}
1174 	return (error);
1175 }
1176 
1177 /*
1178  * As above, but read-only.
1179  */
1180 int
1181 sysctl_rdstring(void *oldp, size_t *oldlenp, void *newp, const char *str)
1182 {
1183 	int error = 0, err2 = 0;
1184 	size_t len;
1185 
1186 	if (newp)
1187 		return (EPERM);
1188 
1189 	SYSCTL_STRING_CORE(oldp, oldlenp, str);
1190 
1191 	return (error);
1192 }
1193 
1194 /*
1195  * Validate parameters and get old / set new parameters
1196  * for a structure oriented sysctl function.
1197  */
1198 int
1199 sysctl_struct(void *oldp, size_t *oldlenp, void *newp, size_t newlen, void *sp,
1200     size_t len)
1201 {
1202 	int error = 0;
1203 
1204 	SYSCTL_SCALAR_NEWPCHECK_LEN(newp, newlen, len)
1205 	SYSCTL_SCALAR_CORE_LEN(oldp, oldlenp, sp, len)
1206 	SYSCTL_SCALAR_NEWPCOP_LEN(newp, sp, len)
1207 
1208 	return (error);
1209 }
1210 
1211 /*
1212  * Validate parameters and get old parameters
1213  * for a structure oriented sysctl function.
1214  */
1215 int
1216 sysctl_rdstruct(void *oldp, size_t *oldlenp, void *newp, const void *sp,
1217     size_t len)
1218 {
1219 	int error = 0;
1220 
1221 	if (newp)
1222 		return (EPERM);
1223 
1224 	SYSCTL_SCALAR_CORE_LEN(oldp, oldlenp, sp, len)
1225 
1226 	return (error);
1227 }
1228 
1229 /*
1230  * As above, but can return a truncated result.
1231  */
1232 int
1233 sysctl_rdminstruct(void *oldp, size_t *oldlenp, void *newp, const void *sp,
1234     size_t len)
1235 {
1236 	int error = 0;
1237 
1238 	if (newp)
1239 		return (EPERM);
1240 
1241 	len = min(*oldlenp, len);
1242 	SYSCTL_SCALAR_CORE_LEN(oldp, oldlenp, sp, len)
1243 
1244 	return (error);
1245 }
1246 
1247 /*
1248  * Get file structures.
1249  */
1250 static int
1251 sysctl_file(void *vwhere, size_t *sizep)
1252 {
1253 	int error;
1254 	size_t buflen;
1255 	struct file *fp;
1256 	char *start, *where;
1257 
1258 	start = where = vwhere;
1259 	buflen = *sizep;
1260 	if (where == NULL) {
1261 		/*
1262 		 * overestimate by 10 files
1263 		 */
1264 		*sizep = sizeof(filehead) + (nfiles + 10) * sizeof(struct file);
1265 		return (0);
1266 	}
1267 
1268 	/*
1269 	 * first copyout filehead
1270 	 */
1271 	if (buflen < sizeof(filehead)) {
1272 		*sizep = 0;
1273 		return (0);
1274 	}
1275 	error = copyout((caddr_t)&filehead, where, sizeof(filehead));
1276 	if (error)
1277 		return (error);
1278 	buflen -= sizeof(filehead);
1279 	where += sizeof(filehead);
1280 
1281 	/*
1282 	 * followed by an array of file structures
1283 	 */
1284 	LIST_FOREACH(fp, &filehead, f_list) {
1285 		if (buflen < sizeof(struct file)) {
1286 			*sizep = where - start;
1287 			return (ENOMEM);
1288 		}
1289 		error = copyout((caddr_t)fp, where, sizeof(struct file));
1290 		if (error)
1291 			return (error);
1292 		buflen -= sizeof(struct file);
1293 		where += sizeof(struct file);
1294 	}
1295 	*sizep = where - start;
1296 	return (0);
1297 }
1298 
1299 #if defined(SYSVMSG) || defined(SYSVSEM) || defined(SYSVSHM)
1300 #define	FILL_PERM(src, dst) do { \
1301 	(dst)._key = (src)._key; \
1302 	(dst).uid = (src).uid; \
1303 	(dst).gid = (src).gid; \
1304 	(dst).cuid = (src).cuid; \
1305 	(dst).cgid = (src).cgid; \
1306 	(dst).mode = (src).mode; \
1307 	(dst)._seq = (src)._seq; \
1308 } while (/*CONSTCOND*/ 0);
1309 #define	FILL_MSG(src, dst) do { \
1310 	FILL_PERM((src).msg_perm, (dst).msg_perm); \
1311 	(dst).msg_qnum = (src).msg_qnum; \
1312 	(dst).msg_qbytes = (src).msg_qbytes; \
1313 	(dst)._msg_cbytes = (src)._msg_cbytes; \
1314 	(dst).msg_lspid = (src).msg_lspid; \
1315 	(dst).msg_lrpid = (src).msg_lrpid; \
1316 	(dst).msg_stime = (src).msg_stime; \
1317 	(dst).msg_rtime = (src).msg_rtime; \
1318 	(dst).msg_ctime = (src).msg_ctime; \
1319 } while (/*CONSTCOND*/ 0)
1320 #define	FILL_SEM(src, dst) do { \
1321 	FILL_PERM((src).sem_perm, (dst).sem_perm); \
1322 	(dst).sem_nsems = (src).sem_nsems; \
1323 	(dst).sem_otime = (src).sem_otime; \
1324 	(dst).sem_ctime = (src).sem_ctime; \
1325 } while (/*CONSTCOND*/ 0)
1326 #define	FILL_SHM(src, dst) do { \
1327 	FILL_PERM((src).shm_perm, (dst).shm_perm); \
1328 	(dst).shm_segsz = (src).shm_segsz; \
1329 	(dst).shm_lpid = (src).shm_lpid; \
1330 	(dst).shm_cpid = (src).shm_cpid; \
1331 	(dst).shm_atime = (src).shm_atime; \
1332 	(dst).shm_dtime = (src).shm_dtime; \
1333 	(dst).shm_ctime = (src).shm_ctime; \
1334 	(dst).shm_nattch = (src).shm_nattch; \
1335 } while (/*CONSTCOND*/ 0)
1336 
1337 static int
1338 sysctl_sysvipc(int *name, u_int namelen, void *where, size_t *sizep)
1339 {
1340 #ifdef SYSVMSG
1341 	struct msg_sysctl_info *msgsi = NULL;
1342 #endif
1343 #ifdef SYSVSEM
1344 	struct sem_sysctl_info *semsi = NULL;
1345 #endif
1346 #ifdef SYSVSHM
1347 	struct shm_sysctl_info *shmsi = NULL;
1348 #endif
1349 	size_t infosize, dssize, tsize, buflen;
1350 	void *buf = NULL;
1351 	char *start;
1352 	int32_t nds;
1353 	int i, error, ret;
1354 
1355 	if (namelen != 1)
1356 		return (EINVAL);
1357 
1358 	start = where;
1359 	buflen = *sizep;
1360 
1361 	switch (*name) {
1362 	case KERN_SYSVIPC_MSG_INFO:
1363 #ifdef SYSVMSG
1364 		infosize = sizeof(msgsi->msginfo);
1365 		nds = msginfo.msgmni;
1366 		dssize = sizeof(msgsi->msgids[0]);
1367 		break;
1368 #else
1369 		return (EINVAL);
1370 #endif
1371 	case KERN_SYSVIPC_SEM_INFO:
1372 #ifdef SYSVSEM
1373 		infosize = sizeof(semsi->seminfo);
1374 		nds = seminfo.semmni;
1375 		dssize = sizeof(semsi->semids[0]);
1376 		break;
1377 #else
1378 		return (EINVAL);
1379 #endif
1380 	case KERN_SYSVIPC_SHM_INFO:
1381 #ifdef SYSVSHM
1382 		infosize = sizeof(shmsi->shminfo);
1383 		nds = shminfo.shmmni;
1384 		dssize = sizeof(shmsi->shmids[0]);
1385 		break;
1386 #else
1387 		return (EINVAL);
1388 #endif
1389 	default:
1390 		return (EINVAL);
1391 	}
1392 	/*
1393 	 * Round infosize to 64 bit boundary if requesting more than just
1394 	 * the info structure or getting the total data size.
1395 	 */
1396 	if (where == NULL || *sizep > infosize)
1397 		infosize = ((infosize + 7) / 8) * 8;
1398 	tsize = infosize + nds * dssize;
1399 
1400 	/* Return just the total size required. */
1401 	if (where == NULL) {
1402 		*sizep = tsize;
1403 		return (0);
1404 	}
1405 
1406 	/* Not enough room for even the info struct. */
1407 	if (buflen < infosize) {
1408 		*sizep = 0;
1409 		return (ENOMEM);
1410 	}
1411 	buf = malloc(min(tsize, buflen), M_TEMP, M_WAITOK);
1412 	memset(buf, 0, min(tsize, buflen));
1413 
1414 	switch (*name) {
1415 #ifdef SYSVMSG
1416 	case KERN_SYSVIPC_MSG_INFO:
1417 		msgsi = (struct msg_sysctl_info *)buf;
1418 		msgsi->msginfo = msginfo;
1419 		break;
1420 #endif
1421 #ifdef SYSVSEM
1422 	case KERN_SYSVIPC_SEM_INFO:
1423 		semsi = (struct sem_sysctl_info *)buf;
1424 		semsi->seminfo = seminfo;
1425 		break;
1426 #endif
1427 #ifdef SYSVSHM
1428 	case KERN_SYSVIPC_SHM_INFO:
1429 		shmsi = (struct shm_sysctl_info *)buf;
1430 		shmsi->shminfo = shminfo;
1431 		break;
1432 #endif
1433 	}
1434 	buflen -= infosize;
1435 
1436 	ret = 0;
1437 	if (buflen > 0) {
1438 		/* Fill in the IPC data structures.  */
1439 		for (i = 0; i < nds; i++) {
1440 			if (buflen < dssize) {
1441 				ret = ENOMEM;
1442 				break;
1443 			}
1444 			switch (*name) {
1445 #ifdef SYSVMSG
1446 			case KERN_SYSVIPC_MSG_INFO:
1447 				FILL_MSG(msqids[i], msgsi->msgids[i]);
1448 				break;
1449 #endif
1450 #ifdef SYSVSEM
1451 			case KERN_SYSVIPC_SEM_INFO:
1452 				FILL_SEM(sema[i], semsi->semids[i]);
1453 				break;
1454 #endif
1455 #ifdef SYSVSHM
1456 			case KERN_SYSVIPC_SHM_INFO:
1457 				FILL_SHM(shmsegs[i], shmsi->shmids[i]);
1458 				break;
1459 #endif
1460 			}
1461 			buflen -= dssize;
1462 		}
1463 	}
1464 	*sizep -= buflen;
1465 	error = copyout(buf, start, *sizep);
1466 	/* If copyout succeeded, use return code set earlier. */
1467 	if (error == 0)
1468 		error = ret;
1469 	if (buf)
1470 		free(buf, M_TEMP);
1471 	return (error);
1472 }
1473 #endif /* SYSVMSG || SYSVSEM || SYSVSHM */
1474 
1475 static int
1476 sysctl_msgbuf(void *vwhere, size_t *sizep)
1477 {
1478 	char *where = vwhere;
1479 	size_t len, maxlen = *sizep;
1480 	long beg, end;
1481 	int error;
1482 
1483 	/*
1484 	 * deal with cases where the message buffer has
1485 	 * become corrupted.
1486 	 */
1487 	if (!msgbufenabled || msgbufp->msg_magic != MSG_MAGIC) {
1488 		msgbufenabled = 0;
1489 		return (ENXIO);
1490 	}
1491 
1492 	if (where == NULL) {
1493 		/* always return full buffer size */
1494 		*sizep = msgbufp->msg_bufs;
1495 		return (0);
1496 	}
1497 
1498 	error = 0;
1499 	maxlen = min(msgbufp->msg_bufs, maxlen);
1500 
1501 	/*
1502 	 * First, copy from the write pointer to the end of
1503 	 * message buffer.
1504 	 */
1505 	beg = msgbufp->msg_bufx;
1506 	end = msgbufp->msg_bufs;
1507 	while (maxlen > 0) {
1508 		len = min(end - beg, maxlen);
1509 		if (len == 0)
1510 			break;
1511 		error = copyout(&msgbufp->msg_bufc[beg], where, len);
1512 		if (error)
1513 			break;
1514 		where += len;
1515 		maxlen -= len;
1516 
1517 		/*
1518 		 * ... then, copy from the beginning of message buffer to
1519 		 * the write pointer.
1520 		 */
1521 		beg = 0;
1522 		end = msgbufp->msg_bufx;
1523 	}
1524 	return (error);
1525 }
1526 
1527 /*
1528  * try over estimating by 5 procs
1529  */
1530 #define KERN_PROCSLOP	(5 * sizeof(struct kinfo_proc))
1531 
1532 static int
1533 sysctl_doeproc(int *name, u_int namelen, void *vwhere, size_t *sizep)
1534 {
1535 	struct eproc eproc;
1536 	struct kinfo_proc2 kproc2;
1537 	struct kinfo_proc *dp;
1538 	struct proc *p;
1539 	const struct proclist_desc *pd;
1540 	char *where, *dp2;
1541 	int type, op, arg;
1542 	u_int elem_size, elem_count;
1543 	size_t buflen, needed;
1544 	int error;
1545 
1546 	dp = vwhere;
1547 	dp2 = where = vwhere;
1548 	buflen = where != NULL ? *sizep : 0;
1549 	error = 0;
1550 	needed = 0;
1551 	type = name[0];
1552 
1553 	if (type == KERN_PROC) {
1554 		if (namelen != 3 && !(namelen == 2 && name[1] == KERN_PROC_ALL))
1555 			return (EINVAL);
1556 		op = name[1];
1557 		if (op != KERN_PROC_ALL)
1558 			arg = name[2];
1559 		else
1560 			arg = 0;		/* Quell compiler warning */
1561 		elem_size = elem_count = 0;	/* Ditto */
1562 	} else {
1563 		if (namelen != 5)
1564 			return (EINVAL);
1565 		op = name[1];
1566 		arg = name[2];
1567 		elem_size = name[3];
1568 		elem_count = name[4];
1569 	}
1570 
1571 	proclist_lock_read();
1572 
1573 	pd = proclists;
1574 again:
1575 	for (p = LIST_FIRST(pd->pd_list); p != NULL; p = LIST_NEXT(p, p_list)) {
1576 		/*
1577 		 * Skip embryonic processes.
1578 		 */
1579 		if (p->p_stat == SIDL)
1580 			continue;
1581 		/*
1582 		 * TODO - make more efficient (see notes below).
1583 		 * do by session.
1584 		 */
1585 		switch (op) {
1586 
1587 		case KERN_PROC_PID:
1588 			/* could do this with just a lookup */
1589 			if (p->p_pid != (pid_t)arg)
1590 				continue;
1591 			break;
1592 
1593 		case KERN_PROC_PGRP:
1594 			/* could do this by traversing pgrp */
1595 			if (p->p_pgrp->pg_id != (pid_t)arg)
1596 				continue;
1597 			break;
1598 
1599 		case KERN_PROC_SESSION:
1600 			if (p->p_session->s_sid != (pid_t)arg)
1601 				continue;
1602 			break;
1603 
1604 		case KERN_PROC_TTY:
1605 			if (arg == (int) KERN_PROC_TTY_REVOKE) {
1606 				if ((p->p_flag & P_CONTROLT) == 0 ||
1607 				    p->p_session->s_ttyp == NULL ||
1608 				    p->p_session->s_ttyvp != NULL)
1609 					continue;
1610 			} else if ((p->p_flag & P_CONTROLT) == 0 ||
1611 			    p->p_session->s_ttyp == NULL) {
1612 				if ((dev_t)arg != KERN_PROC_TTY_NODEV)
1613 					continue;
1614 			} else if (p->p_session->s_ttyp->t_dev != (dev_t)arg)
1615 				continue;
1616 			break;
1617 
1618 		case KERN_PROC_UID:
1619 			if (p->p_ucred->cr_uid != (uid_t)arg)
1620 				continue;
1621 			break;
1622 
1623 		case KERN_PROC_RUID:
1624 			if (p->p_cred->p_ruid != (uid_t)arg)
1625 				continue;
1626 			break;
1627 
1628 		case KERN_PROC_GID:
1629 			if (p->p_ucred->cr_gid != (uid_t)arg)
1630 				continue;
1631 			break;
1632 
1633 		case KERN_PROC_RGID:
1634 			if (p->p_cred->p_rgid != (uid_t)arg)
1635 				continue;
1636 			break;
1637 
1638 		case KERN_PROC_ALL:
1639 			/* allow everything */
1640 			break;
1641 
1642 		default:
1643 			error = EINVAL;
1644 			goto cleanup;
1645 		}
1646 		if (type == KERN_PROC) {
1647 			if (buflen >= sizeof(struct kinfo_proc)) {
1648 				fill_eproc(p, &eproc);
1649 				error = copyout((caddr_t)p, &dp->kp_proc,
1650 				    sizeof(struct proc));
1651 				if (error)
1652 					goto cleanup;
1653 				error = copyout((caddr_t)&eproc, &dp->kp_eproc,
1654 				    sizeof(eproc));
1655 				if (error)
1656 					goto cleanup;
1657 				dp++;
1658 				buflen -= sizeof(struct kinfo_proc);
1659 			}
1660 			needed += sizeof(struct kinfo_proc);
1661 		} else { /* KERN_PROC2 */
1662 			if (buflen >= elem_size && elem_count > 0) {
1663 				fill_kproc2(p, &kproc2);
1664 				/*
1665 				 * Copy out elem_size, but not larger than
1666 				 * the size of a struct kinfo_proc2.
1667 				 */
1668 				error = copyout(&kproc2, dp2,
1669 				    min(sizeof(kproc2), elem_size));
1670 				if (error)
1671 					goto cleanup;
1672 				dp2 += elem_size;
1673 				buflen -= elem_size;
1674 				elem_count--;
1675 			}
1676 			needed += elem_size;
1677 		}
1678 	}
1679 	pd++;
1680 	if (pd->pd_list != NULL)
1681 		goto again;
1682 	proclist_unlock_read();
1683 
1684 	if (where != NULL) {
1685 		if (type == KERN_PROC)
1686 			*sizep = (caddr_t)dp - where;
1687 		else
1688 			*sizep = dp2 - where;
1689 		if (needed > *sizep)
1690 			return (ENOMEM);
1691 	} else {
1692 		needed += KERN_PROCSLOP;
1693 		*sizep = needed;
1694 	}
1695 	return (0);
1696  cleanup:
1697 	proclist_unlock_read();
1698 	return (error);
1699 }
1700 
1701 
1702 /*
1703  * try over estimating by 5 LWPs
1704  */
1705 #define KERN_LWPSLOP	(5 * sizeof(struct kinfo_lwp))
1706 
1707 static int
1708 sysctl_dolwp(int *name, u_int namelen, void *vwhere, size_t *sizep)
1709 {
1710 	struct kinfo_lwp klwp;
1711 	struct proc *p;
1712 	struct lwp *l;
1713 	char *where, *dp;
1714 	int type, pid, elem_size, elem_count;
1715 	int buflen, needed, error;
1716 
1717 	dp = where = vwhere;
1718 	buflen = where != NULL ? *sizep : 0;
1719 	error = needed = 0;
1720 	type = name[0];
1721 
1722 	if (namelen != 4)
1723 		return (EINVAL);
1724 	pid = name[1];
1725 	elem_size = name[2];
1726 	elem_count = name[3];
1727 
1728 	p = pfind(pid);
1729 	if (p == NULL)
1730 		return (ESRCH);
1731 	LIST_FOREACH(l, &p->p_lwps, l_sibling) {
1732 		if (buflen >= elem_size && elem_count > 0) {
1733 			fill_lwp(l, &klwp);
1734 			/*
1735 			 * Copy out elem_size, but not larger than
1736 			 * the size of a struct kinfo_proc2.
1737 			 */
1738 			error = copyout(&klwp, dp,
1739 			    min(sizeof(klwp), elem_size));
1740 			if (error)
1741 				goto cleanup;
1742 			dp += elem_size;
1743 			buflen -= elem_size;
1744 			elem_count--;
1745 		}
1746 		needed += elem_size;
1747 	}
1748 
1749 	if (where != NULL) {
1750 		*sizep = dp - where;
1751 		if (needed > *sizep)
1752 			return (ENOMEM);
1753 	} else {
1754 		needed += KERN_PROCSLOP;
1755 		*sizep = needed;
1756 	}
1757 	return (0);
1758  cleanup:
1759 	return (error);
1760 }
1761 
1762 /*
1763  * Fill in an eproc structure for the specified process.
1764  */
1765 void
1766 fill_eproc(struct proc *p, struct eproc *ep)
1767 {
1768 	struct tty *tp;
1769 	struct lwp *l;
1770 
1771 	ep->e_paddr = p;
1772 	ep->e_sess = p->p_session;
1773 	ep->e_pcred = *p->p_cred;
1774 	ep->e_ucred = *p->p_ucred;
1775 	if (p->p_stat == SIDL || P_ZOMBIE(p)) {
1776 		ep->e_vm.vm_rssize = 0;
1777 		ep->e_vm.vm_tsize = 0;
1778 		ep->e_vm.vm_dsize = 0;
1779 		ep->e_vm.vm_ssize = 0;
1780 		/* ep->e_vm.vm_pmap = XXX; */
1781 	} else {
1782 		struct vmspace *vm = p->p_vmspace;
1783 
1784 		ep->e_vm.vm_rssize = vm_resident_count(vm);
1785 		ep->e_vm.vm_tsize = vm->vm_tsize;
1786 		ep->e_vm.vm_dsize = vm->vm_dsize;
1787 		ep->e_vm.vm_ssize = vm->vm_ssize;
1788 
1789 		/* Pick a "representative" LWP */
1790 		l = proc_representative_lwp(p);
1791 
1792 		if (l->l_wmesg)
1793 			strncpy(ep->e_wmesg, l->l_wmesg, WMESGLEN);
1794 	}
1795 	if (p->p_pptr)
1796 		ep->e_ppid = p->p_pptr->p_pid;
1797 	else
1798 		ep->e_ppid = 0;
1799 	ep->e_pgid = p->p_pgrp->pg_id;
1800 	ep->e_sid = ep->e_sess->s_sid;
1801 	ep->e_jobc = p->p_pgrp->pg_jobc;
1802 	if ((p->p_flag & P_CONTROLT) &&
1803 	    (tp = ep->e_sess->s_ttyp)) {
1804 		ep->e_tdev = tp->t_dev;
1805 		ep->e_tpgid = tp->t_pgrp ? tp->t_pgrp->pg_id : NO_PGID;
1806 		ep->e_tsess = tp->t_session;
1807 	} else
1808 		ep->e_tdev = NODEV;
1809 
1810 	ep->e_xsize = ep->e_xrssize = 0;
1811 	ep->e_xccount = ep->e_xswrss = 0;
1812 	ep->e_flag = ep->e_sess->s_ttyvp ? EPROC_CTTY : 0;
1813 	if (SESS_LEADER(p))
1814 		ep->e_flag |= EPROC_SLEADER;
1815 	strncpy(ep->e_login, ep->e_sess->s_login, MAXLOGNAME);
1816 }
1817 
1818 /*
1819  * Fill in an eproc structure for the specified process.
1820  */
1821 static void
1822 fill_kproc2(struct proc *p, struct kinfo_proc2 *ki)
1823 {
1824 	struct tty *tp;
1825 	struct lwp *l;
1826 	struct timeval ut, st;
1827 
1828 	memset(ki, 0, sizeof(*ki));
1829 
1830 	ki->p_paddr = PTRTOINT64(p);
1831 	ki->p_fd = PTRTOINT64(p->p_fd);
1832 	ki->p_cwdi = PTRTOINT64(p->p_cwdi);
1833 	ki->p_stats = PTRTOINT64(p->p_stats);
1834 	ki->p_limit = PTRTOINT64(p->p_limit);
1835 	ki->p_vmspace = PTRTOINT64(p->p_vmspace);
1836 	ki->p_sigacts = PTRTOINT64(p->p_sigacts);
1837 	ki->p_sess = PTRTOINT64(p->p_session);
1838 	ki->p_tsess = 0;	/* may be changed if controlling tty below */
1839 	ki->p_ru = PTRTOINT64(p->p_ru);
1840 
1841 	ki->p_eflag = 0;
1842 	ki->p_exitsig = p->p_exitsig;
1843 	ki->p_flag = p->p_flag;
1844 
1845 	ki->p_pid = p->p_pid;
1846 	if (p->p_pptr)
1847 		ki->p_ppid = p->p_pptr->p_pid;
1848 	else
1849 		ki->p_ppid = 0;
1850 	ki->p_sid = p->p_session->s_sid;
1851 	ki->p__pgid = p->p_pgrp->pg_id;
1852 
1853 	ki->p_tpgid = NO_PGID;	/* may be changed if controlling tty below */
1854 
1855 	ki->p_uid = p->p_ucred->cr_uid;
1856 	ki->p_ruid = p->p_cred->p_ruid;
1857 	ki->p_gid = p->p_ucred->cr_gid;
1858 	ki->p_rgid = p->p_cred->p_rgid;
1859 	ki->p_svuid = p->p_cred->p_svuid;
1860 	ki->p_svgid = p->p_cred->p_svgid;
1861 
1862 	memcpy(ki->p_groups, p->p_cred->pc_ucred->cr_groups,
1863 	    min(sizeof(ki->p_groups), sizeof(p->p_cred->pc_ucred->cr_groups)));
1864 	ki->p_ngroups = p->p_cred->pc_ucred->cr_ngroups;
1865 
1866 	ki->p_jobc = p->p_pgrp->pg_jobc;
1867 	if ((p->p_flag & P_CONTROLT) && (tp = p->p_session->s_ttyp)) {
1868 		ki->p_tdev = tp->t_dev;
1869 		ki->p_tpgid = tp->t_pgrp ? tp->t_pgrp->pg_id : NO_PGID;
1870 		ki->p_tsess = PTRTOINT64(tp->t_session);
1871 	} else {
1872 		ki->p_tdev = NODEV;
1873 	}
1874 
1875 	ki->p_estcpu = p->p_estcpu;
1876 	ki->p_rtime_sec = p->p_rtime.tv_sec;
1877 	ki->p_rtime_usec = p->p_rtime.tv_usec;
1878 	ki->p_cpticks = p->p_cpticks;
1879 	ki->p_pctcpu = p->p_pctcpu;
1880 
1881 	ki->p_uticks = p->p_uticks;
1882 	ki->p_sticks = p->p_sticks;
1883 	ki->p_iticks = p->p_iticks;
1884 
1885 	ki->p_tracep = PTRTOINT64(p->p_tracep);
1886 	ki->p_traceflag = p->p_traceflag;
1887 
1888 
1889 	memcpy(&ki->p_siglist, &p->p_sigctx.ps_siglist, sizeof(ki_sigset_t));
1890 	memcpy(&ki->p_sigmask, &p->p_sigctx.ps_sigmask, sizeof(ki_sigset_t));
1891 	memcpy(&ki->p_sigignore, &p->p_sigctx.ps_sigignore,sizeof(ki_sigset_t));
1892 	memcpy(&ki->p_sigcatch, &p->p_sigctx.ps_sigcatch, sizeof(ki_sigset_t));
1893 
1894 	ki->p_stat = p->p_stat; /* Will likely be overridden by LWP status */
1895 	ki->p_realstat = p->p_stat;
1896 	ki->p_nice = p->p_nice;
1897 
1898 	ki->p_xstat = p->p_xstat;
1899 	ki->p_acflag = p->p_acflag;
1900 
1901 	strncpy(ki->p_comm, p->p_comm,
1902 	    min(sizeof(ki->p_comm), sizeof(p->p_comm)));
1903 
1904 	strncpy(ki->p_login, p->p_session->s_login,
1905 	    min(sizeof ki->p_login - 1, sizeof p->p_session->s_login));
1906 
1907 	ki->p_nlwps = p->p_nlwps;
1908 	ki->p_nrlwps = p->p_nrlwps;
1909 	ki->p_realflag = p->p_flag;
1910 
1911 	if (p->p_stat == SIDL || P_ZOMBIE(p)) {
1912 		ki->p_vm_rssize = 0;
1913 		ki->p_vm_tsize = 0;
1914 		ki->p_vm_dsize = 0;
1915 		ki->p_vm_ssize = 0;
1916 		l = NULL;
1917 	} else {
1918 		struct vmspace *vm = p->p_vmspace;
1919 
1920 		ki->p_vm_rssize = vm_resident_count(vm);
1921 		ki->p_vm_tsize = vm->vm_tsize;
1922 		ki->p_vm_dsize = vm->vm_dsize;
1923 		ki->p_vm_ssize = vm->vm_ssize;
1924 
1925 		/* Pick a "representative" LWP */
1926 		l = proc_representative_lwp(p);
1927 		ki->p_forw = PTRTOINT64(l->l_forw);
1928 		ki->p_back = PTRTOINT64(l->l_back);
1929 		ki->p_addr = PTRTOINT64(l->l_addr);
1930 		ki->p_stat = l->l_stat;
1931 		ki->p_flag |= l->l_flag;
1932 		ki->p_swtime = l->l_swtime;
1933 		ki->p_slptime = l->l_slptime;
1934 		if (l->l_stat == LSONPROC) {
1935 			KDASSERT(l->l_cpu != NULL);
1936 			ki->p_schedflags = l->l_cpu->ci_schedstate.spc_flags;
1937 		} else
1938 			ki->p_schedflags = 0;
1939 		ki->p_holdcnt = l->l_holdcnt;
1940 		ki->p_priority = l->l_priority;
1941 		ki->p_usrpri = l->l_usrpri;
1942 		if (l->l_wmesg)
1943 			strncpy(ki->p_wmesg, l->l_wmesg, sizeof(ki->p_wmesg));
1944 		ki->p_wchan = PTRTOINT64(l->l_wchan);
1945 
1946 	}
1947 
1948 	if (p->p_session->s_ttyvp)
1949 		ki->p_eflag |= EPROC_CTTY;
1950 	if (SESS_LEADER(p))
1951 		ki->p_eflag |= EPROC_SLEADER;
1952 
1953 	/* XXX Is this double check necessary? */
1954 	if (P_ZOMBIE(p)) {
1955 		ki->p_uvalid = 0;
1956 	} else {
1957 		ki->p_uvalid = 1;
1958 
1959 		ki->p_ustart_sec = p->p_stats->p_start.tv_sec;
1960 		ki->p_ustart_usec = p->p_stats->p_start.tv_usec;
1961 
1962 		calcru(p, &ut, &st, 0);
1963 		ki->p_uutime_sec = ut.tv_sec;
1964 		ki->p_uutime_usec = ut.tv_usec;
1965 		ki->p_ustime_sec = st.tv_sec;
1966 		ki->p_ustime_usec = st.tv_usec;
1967 
1968 		ki->p_uru_maxrss = p->p_stats->p_ru.ru_maxrss;
1969 		ki->p_uru_ixrss = p->p_stats->p_ru.ru_ixrss;
1970 		ki->p_uru_idrss = p->p_stats->p_ru.ru_idrss;
1971 		ki->p_uru_isrss = p->p_stats->p_ru.ru_isrss;
1972 		ki->p_uru_minflt = p->p_stats->p_ru.ru_minflt;
1973 		ki->p_uru_majflt = p->p_stats->p_ru.ru_majflt;
1974 		ki->p_uru_nswap = p->p_stats->p_ru.ru_nswap;
1975 		ki->p_uru_inblock = p->p_stats->p_ru.ru_inblock;
1976 		ki->p_uru_oublock = p->p_stats->p_ru.ru_oublock;
1977 		ki->p_uru_msgsnd = p->p_stats->p_ru.ru_msgsnd;
1978 		ki->p_uru_msgrcv = p->p_stats->p_ru.ru_msgrcv;
1979 		ki->p_uru_nsignals = p->p_stats->p_ru.ru_nsignals;
1980 		ki->p_uru_nvcsw = p->p_stats->p_ru.ru_nvcsw;
1981 		ki->p_uru_nivcsw = p->p_stats->p_ru.ru_nivcsw;
1982 
1983 		timeradd(&p->p_stats->p_cru.ru_utime,
1984 			 &p->p_stats->p_cru.ru_stime, &ut);
1985 		ki->p_uctime_sec = ut.tv_sec;
1986 		ki->p_uctime_usec = ut.tv_usec;
1987 	}
1988 #ifdef MULTIPROCESSOR
1989 	if (l && l->l_cpu != NULL)
1990 		ki->p_cpuid = l->l_cpu->ci_cpuid;
1991 	else
1992 #endif
1993 		ki->p_cpuid = KI_NOCPU;
1994 
1995 }
1996 
1997 /*
1998  * Fill in a kinfo_lwp structure for the specified lwp.
1999  */
2000 static void
2001 fill_lwp(struct lwp *l, struct kinfo_lwp *kl)
2002 {
2003 
2004 	kl->l_forw = PTRTOINT64(l->l_forw);
2005 	kl->l_back = PTRTOINT64(l->l_back);
2006 	kl->l_laddr = PTRTOINT64(l);
2007 	kl->l_addr = PTRTOINT64(l->l_addr);
2008 	kl->l_stat = l->l_stat;
2009 	kl->l_lid = l->l_lid;
2010 	kl->l_flag = l->l_flag;
2011 
2012 	kl->l_swtime = l->l_swtime;
2013 	kl->l_slptime = l->l_slptime;
2014 	if (l->l_stat == LSONPROC) {
2015 		KDASSERT(l->l_cpu != NULL);
2016 		kl->l_schedflags = l->l_cpu->ci_schedstate.spc_flags;
2017 	} else
2018 		kl->l_schedflags = 0;
2019 	kl->l_holdcnt = l->l_holdcnt;
2020 	kl->l_priority = l->l_priority;
2021 	kl->l_usrpri = l->l_usrpri;
2022 	if (l->l_wmesg)
2023 		strncpy(kl->l_wmesg, l->l_wmesg, sizeof(kl->l_wmesg));
2024 	kl->l_wchan = PTRTOINT64(l->l_wchan);
2025 #ifdef MULTIPROCESSOR
2026 	if (l->l_cpu != NULL)
2027 		kl->l_cpuid = l->l_cpu->ci_cpuid;
2028 	else
2029 #endif
2030 		kl->l_cpuid = KI_NOCPU;
2031 }
2032 
2033 int
2034 sysctl_procargs(int *name, u_int namelen, void *where, size_t *sizep,
2035     struct proc *up)
2036 {
2037 	struct ps_strings pss;
2038 	struct proc *p;
2039 	size_t len, upper_bound, xlen, i;
2040 	struct uio auio;
2041 	struct iovec aiov;
2042 	vaddr_t argv;
2043 	pid_t pid;
2044 	int nargv, type, error;
2045 	char *arg;
2046 	char *tmp;
2047 
2048 	if (namelen != 2)
2049 		return (EINVAL);
2050 	pid = name[0];
2051 	type = name[1];
2052 
2053 	switch (type) {
2054 	case KERN_PROC_ARGV:
2055 	case KERN_PROC_NARGV:
2056 	case KERN_PROC_ENV:
2057 	case KERN_PROC_NENV:
2058 		/* ok */
2059 		break;
2060 	default:
2061 		return (EINVAL);
2062 	}
2063 
2064 	/* check pid */
2065 	if ((p = pfind(pid)) == NULL)
2066 		return (EINVAL);
2067 
2068 	/* only root or same user change look at the environment */
2069 	if (type == KERN_PROC_ENV || type == KERN_PROC_NENV) {
2070 		if (up->p_ucred->cr_uid != 0) {
2071 			if (up->p_cred->p_ruid != p->p_cred->p_ruid ||
2072 			    up->p_cred->p_ruid != p->p_cred->p_svuid)
2073 				return (EPERM);
2074 		}
2075 	}
2076 
2077 	if (sizep != NULL && where == NULL) {
2078 		if (type == KERN_PROC_NARGV || type == KERN_PROC_NENV)
2079 			*sizep = sizeof (int);
2080 		else
2081 			*sizep = ARG_MAX;	/* XXX XXX XXX */
2082 		return (0);
2083 	}
2084 	if (where == NULL || sizep == NULL)
2085 		return (EINVAL);
2086 
2087 	/*
2088 	 * Zombies don't have a stack, so we can't read their psstrings.
2089 	 * System processes also don't have a user stack.
2090 	 */
2091 	if (P_ZOMBIE(p) || (p->p_flag & P_SYSTEM) != 0)
2092 		return (EINVAL);
2093 
2094 	/*
2095 	 * Lock the process down in memory.
2096 	 */
2097 	/* XXXCDC: how should locking work here? */
2098 	if ((p->p_flag & P_WEXIT) || (p->p_vmspace->vm_refcnt < 1))
2099 		return (EFAULT);
2100 
2101 	p->p_vmspace->vm_refcnt++;	/* XXX */
2102 
2103 	/*
2104 	 * Allocate a temporary buffer to hold the arguments.
2105 	 */
2106 	arg = malloc(PAGE_SIZE, M_TEMP, M_WAITOK);
2107 
2108 	/*
2109 	 * Read in the ps_strings structure.
2110 	 */
2111 	aiov.iov_base = &pss;
2112 	aiov.iov_len = sizeof(pss);
2113 	auio.uio_iov = &aiov;
2114 	auio.uio_iovcnt = 1;
2115 	auio.uio_offset = (vaddr_t)p->p_psstr;
2116 	auio.uio_resid = sizeof(pss);
2117 	auio.uio_segflg = UIO_SYSSPACE;
2118 	auio.uio_rw = UIO_READ;
2119 	auio.uio_procp = NULL;
2120 	error = uvm_io(&p->p_vmspace->vm_map, &auio);
2121 	if (error)
2122 		goto done;
2123 
2124 	if (type == KERN_PROC_ARGV || type == KERN_PROC_NARGV)
2125 		memcpy(&nargv, (char *)&pss + p->p_psnargv, sizeof(nargv));
2126 	else
2127 		memcpy(&nargv, (char *)&pss + p->p_psnenv, sizeof(nargv));
2128 	if (type == KERN_PROC_NARGV || type == KERN_PROC_NENV) {
2129 		error = copyout(&nargv, where, sizeof(nargv));
2130 		*sizep = sizeof(nargv);
2131 		goto done;
2132 	}
2133 	/*
2134 	 * Now read the address of the argument vector.
2135 	 */
2136 	switch (type) {
2137 	case KERN_PROC_ARGV:
2138 		/* XXX compat32 stuff here */
2139 		memcpy(&tmp, (char *)&pss + p->p_psargv, sizeof(tmp));
2140 		break;
2141 	case KERN_PROC_ENV:
2142 		memcpy(&tmp, (char *)&pss + p->p_psenv, sizeof(tmp));
2143 		break;
2144 	default:
2145 		return (EINVAL);
2146 	}
2147 	auio.uio_offset = (off_t)(long)tmp;
2148 	aiov.iov_base = &argv;
2149 	aiov.iov_len = sizeof(argv);
2150 	auio.uio_iov = &aiov;
2151 	auio.uio_iovcnt = 1;
2152 	auio.uio_resid = sizeof(argv);
2153 	auio.uio_segflg = UIO_SYSSPACE;
2154 	auio.uio_rw = UIO_READ;
2155 	auio.uio_procp = NULL;
2156 	error = uvm_io(&p->p_vmspace->vm_map, &auio);
2157 	if (error)
2158 		goto done;
2159 
2160 	/*
2161 	 * Now copy in the actual argument vector, one page at a time,
2162 	 * since we don't know how long the vector is (though, we do
2163 	 * know how many NUL-terminated strings are in the vector).
2164 	 */
2165 	len = 0;
2166 	upper_bound = *sizep;
2167 	for (; nargv != 0 && len < upper_bound; len += xlen) {
2168 		aiov.iov_base = arg;
2169 		aiov.iov_len = PAGE_SIZE;
2170 		auio.uio_iov = &aiov;
2171 		auio.uio_iovcnt = 1;
2172 		auio.uio_offset = argv + len;
2173 		xlen = PAGE_SIZE - ((argv + len) & PAGE_MASK);
2174 		auio.uio_resid = xlen;
2175 		auio.uio_segflg = UIO_SYSSPACE;
2176 		auio.uio_rw = UIO_READ;
2177 		auio.uio_procp = NULL;
2178 		error = uvm_io(&p->p_vmspace->vm_map, &auio);
2179 		if (error)
2180 			goto done;
2181 
2182 		for (i = 0; i < xlen && nargv != 0; i++) {
2183 			if (arg[i] == '\0')
2184 				nargv--;	/* one full string */
2185 		}
2186 
2187 		/*
2188 		 * Make sure we don't copyout past the end of the user's
2189 		 * buffer.
2190 		 */
2191 		if (len + i > upper_bound)
2192 			i = upper_bound - len;
2193 
2194 		error = copyout(arg, (char *)where + len, i);
2195 		if (error)
2196 			break;
2197 
2198 		if (nargv == 0) {
2199 			len += i;
2200 			break;
2201 		}
2202 	}
2203 	*sizep = len;
2204 
2205 done:
2206 	uvmspace_free(p->p_vmspace);
2207 
2208 	free(arg, M_TEMP);
2209 	return (error);
2210 }
2211 
2212 #if NPTY > 0
2213 int pty_maxptys(int, int);		/* defined in kern/tty_pty.c */
2214 
2215 /*
2216  * Validate parameters and get old / set new parameters
2217  * for pty sysctl function.
2218  */
2219 static int
2220 sysctl_pty(void *oldp, size_t *oldlenp, void *newp, size_t newlen)
2221 {
2222 	int error = 0;
2223 	int oldmax = 0, newmax = 0;
2224 
2225 	/* get current value of maxptys */
2226 	oldmax = pty_maxptys(0, 0);
2227 
2228 	SYSCTL_SCALAR_CORE_TYP(oldp, oldlenp, &oldmax, int)
2229 
2230 	if (!error && newp) {
2231 		SYSCTL_SCALAR_NEWPCHECK_TYP(newp, newlen, int)
2232 		SYSCTL_SCALAR_NEWPCOP_TYP(newp, &newmax, int)
2233 
2234 		if (newmax != pty_maxptys(newmax, (newp != NULL)))
2235 			return (EINVAL);
2236 
2237 	}
2238 
2239 	return (error);
2240 }
2241 #endif /* NPTY > 0 */
2242 
2243 static int
2244 sysctl_dotkstat(int *name, u_int namelen, void *where, size_t *sizep,
2245     void *newp)
2246 {
2247 
2248 	/* all sysctl names at this level are terminal */
2249 	if (namelen != 1)
2250 		return (ENOTDIR);		/* overloaded */
2251 
2252 	switch (name[0]) {
2253 	case KERN_TKSTAT_NIN:
2254 		return (sysctl_rdquad(where, sizep, newp, tk_nin));
2255 	case KERN_TKSTAT_NOUT:
2256 		return (sysctl_rdquad(where, sizep, newp, tk_nout));
2257 	case KERN_TKSTAT_CANCC:
2258 		return (sysctl_rdquad(where, sizep, newp, tk_cancc));
2259 	case KERN_TKSTAT_RAWCC:
2260 		return (sysctl_rdquad(where, sizep, newp, tk_rawcc));
2261 	default:
2262 		return (EOPNOTSUPP);
2263 	}
2264 }
2265