kern_sysctl.c revision 12260 
1/*-
2 * Copyright (c) 1982, 1986, 1989, 1993
3 *	The Regents of the University of California.  All rights reserved.
4 *
5 * This code is derived from software contributed to Berkeley by
6 * Mike Karels at Berkeley Software Design, Inc.
7 *
8 * Redistribution and use in source and binary forms, with or without
9 * modification, are permitted provided that the following conditions
10 * are met:
11 * 1. Redistributions of source code must retain the above copyright
12 *    notice, this list of conditions and the following disclaimer.
13 * 2. Redistributions in binary form must reproduce the above copyright
14 *    notice, this list of conditions and the following disclaimer in the
15 *    documentation and/or other materials provided with the distribution.
16 * 3. All advertising materials mentioning features or use of this software
17 *    must display the following acknowledgement:
18 *	This product includes software developed by the University of
19 *	California, Berkeley and its contributors.
20 * 4. Neither the name of the University nor the names of its contributors
21 *    may be used to endorse or promote products derived from this software
22 *    without specific prior written permission.
23 *
24 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
25 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
26 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
27 * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
28 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
29 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
30 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
31 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
32 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
33 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
34 * SUCH DAMAGE.
35 *
36 *	@(#)kern_sysctl.c	8.4 (Berkeley) 4/14/94
37 * $Id: kern_sysctl.c,v 1.38 1995/11/12 19:51:51 phk Exp $
38 */
39
40/*
41 * sysctl system call.
42 */
43
44#include <sys/param.h>
45#include <sys/systm.h>
46#include <sys/sysproto.h>
47#include <sys/kernel.h>
48#include <sys/malloc.h>
49#include <sys/proc.h>
50#include <sys/file.h>
51#include <sys/vnode.h>
52#include <sys/unistd.h>
53#include <sys/buf.h>
54#include <sys/ioctl.h>
55#include <sys/tty.h>
56#include <sys/conf.h>
57#include <vm/vm.h>
58#include <sys/sysctl.h>
59#include <sys/user.h>
60
61extern struct linker_set sysctl_;
62
63/* BEGIN_MIB */
64SYSCTL_NODE(, 0,	  sysctl, CTLFLAG_RW, 0,
65	"Sysctl internal magic");
66SYSCTL_NODE(, CTL_KERN,	  kern,   CTLFLAG_RW, 0,
67	"High kernel, proc, limits &c");
68SYSCTL_NODE(, CTL_VM,	  vm,     CTLFLAG_RW, 0,
69	"Virtual memory");
70SYSCTL_NODE(, CTL_FS,	  fs,     CTLFLAG_RW, 0,
71	"File system");
72SYSCTL_NODE(, CTL_NET,	  net,    CTLFLAG_RW, 0,
73	"Network, (see socket.h)");
74SYSCTL_NODE(, CTL_DEBUG,  debug,  CTLFLAG_RW, 0,
75	"Debugging");
76SYSCTL_NODE(, CTL_HW,	  hw,     CTLFLAG_RW, 0,
77	"hardware");
78SYSCTL_NODE(, CTL_MACHDEP, machdep, CTLFLAG_RW, 0,
79	"machine dependent");
80SYSCTL_NODE(, CTL_USER,	  user,   CTLFLAG_RW, 0,
81	"user-level");
82
83SYSCTL_STRING(_kern, KERN_OSRELEASE, osrelease, CTLFLAG_RD, osrelease, 0, "");
84
85SYSCTL_INT(_kern, KERN_OSREV, osrevision, CTLFLAG_RD, 0, BSD, "");
86
87SYSCTL_STRING(_kern, KERN_VERSION, version, CTLFLAG_RD, version, 0, "");
88
89SYSCTL_STRING(_kern, KERN_OSTYPE, ostype, CTLFLAG_RD, ostype, 0, "");
90
91extern int osreldate;
92SYSCTL_INT(_kern, KERN_OSRELDATE, osreldate, CTLFLAG_RD, &osreldate, 0, "");
93
94SYSCTL_INT(_kern, KERN_MAXVNODES, maxvnodes, CTLFLAG_RD, &desiredvnodes, 0, "");
95
96SYSCTL_INT(_kern, KERN_MAXPROC, maxproc, CTLFLAG_RD, &maxproc, 0, "");
97
98SYSCTL_INT(_kern, KERN_MAXPROCPERUID, maxprocperuid,
99	CTLFLAG_RD, &maxprocperuid, 0, "");
100
101SYSCTL_INT(_kern, KERN_MAXFILESPERPROC, maxfilesperproc,
102	CTLFLAG_RD, &maxfilesperproc, 0, "");
103
104SYSCTL_INT(_kern, KERN_ARGMAX, argmax, CTLFLAG_RD, 0, ARG_MAX, "");
105
106SYSCTL_INT(_kern, KERN_POSIX1, posix1version, CTLFLAG_RD, 0, _POSIX_VERSION, "");
107
108SYSCTL_INT(_kern, KERN_NGROUPS, ngroups, CTLFLAG_RD, 0, NGROUPS_MAX, "");
109
110SYSCTL_INT(_kern, KERN_JOB_CONTROL, job_control, CTLFLAG_RD, 0, 1, "");
111
112SYSCTL_INT(_kern, KERN_MAXFILES, maxfiles, CTLFLAG_RW, &maxfiles, 0, "");
113
114#ifdef _POSIX_SAVED_IDS
115SYSCTL_INT(_kern, KERN_SAVED_IDS, saved_ids, CTLFLAG_RD, 0, 1, "");
116#else
117SYSCTL_INT(_kern, KERN_SAVED_IDS, saved_ids, CTLFLAG_RD, 0, 0, "");
118#endif
119
120char kernelname[MAXPATHLEN] = "/kernel";	/* XXX bloat */
121
122SYSCTL_STRING(_kern, KERN_BOOTFILE, bootfile,
123	CTLFLAG_RW, kernelname, sizeof kernelname, "");
124
125SYSCTL_STRUCT(_kern, KERN_BOOTTIME, boottime,
126	CTLFLAG_RW, &boottime, timeval, "");
127
128SYSCTL_STRING(_hw, HW_MACHINE, machine, CTLFLAG_RD, machine, 0, "");
129
130SYSCTL_STRING(_hw, HW_MODEL, model, CTLFLAG_RD, cpu_model, 0, "");
131
132SYSCTL_INT(_hw, HW_NCPU, ncpu, CTLFLAG_RD, 0, 1, "");
133
134SYSCTL_INT(_hw, HW_BYTEORDER, byteorder, CTLFLAG_RD, 0, BYTE_ORDER, "");
135
136SYSCTL_INT(_hw, HW_PAGESIZE, pagesize, CTLFLAG_RD, 0, PAGE_SIZE, "");
137
138/* END_MIB */
139
140extern int vfs_update_wakeup;
141extern int vfs_update_interval;
142static int
143sysctl_kern_updateinterval SYSCTL_HANDLER_ARGS
144{
145	int error = sysctl_handle_int(oidp,
146		oidp->oid_arg1, oidp->oid_arg2, req);
147	if (!error)
148		wakeup(&vfs_update_wakeup);
149	return error;
150}
151
152SYSCTL_PROC(_kern, KERN_UPDATEINTERVAL, update, CTLTYPE_INT|CTLFLAG_RW,
153	&vfs_update_interval, 0, sysctl_kern_updateinterval, "");
154
155
156char hostname[MAXHOSTNAMELEN];
157int hostnamelen;
158static int
159sysctl_kern_hostname SYSCTL_HANDLER_ARGS
160{
161	int error = sysctl_handle_string(oidp,
162		oidp->oid_arg1, oidp->oid_arg2, req);
163	if (req->newptr && (error == 0 || error == ENOMEM))
164		hostnamelen = req->newlen;
165	return error;
166}
167
168SYSCTL_PROC(_kern, KERN_HOSTNAME, hostname, CTLTYPE_STRING|CTLFLAG_RW,
169	&hostname, sizeof(hostname), sysctl_kern_hostname, "");
170
171static int
172sysctl_order_cmp(const void *a, const void *b)
173{
174	const struct sysctl_oid **pa, **pb;
175
176	pa = (const struct sysctl_oid **)a;
177	pb = (const struct sysctl_oid **)b;
178	if (*pa == NULL)
179		return (1);
180	if (*pb == NULL)
181		return (-1);
182	return ((*pa)->oid_number - (*pb)->oid_number);
183}
184
185static void
186sysctl_order(void *arg)
187{
188	int j;
189	struct linker_set *l = (struct linker_set *) arg;
190	struct sysctl_oid **oidpp;
191
192	j = l->ls_length;
193	oidpp = (struct sysctl_oid **) l->ls_items;
194	for (; j--; oidpp++) {
195		if (!*oidpp)
196			continue;
197		if ((*oidpp)->oid_arg1 == arg) {
198			*oidpp = 0;
199			continue;
200		}
201		if (((*oidpp)->oid_kind & CTLTYPE) == CTLTYPE_NODE)
202			if (!(*oidpp)->oid_handler)
203				sysctl_order((*oidpp)->oid_arg1);
204	}
205	qsort(l->ls_items, l->ls_length, sizeof l->ls_items[0],
206		sysctl_order_cmp);
207}
208
209SYSINIT(sysctl, SI_SUB_KMEM, SI_ORDER_ANY, sysctl_order, &sysctl_);
210
211static void
212sysctl_sysctl_debug_dump_node(struct linker_set *l, int i)
213{
214	int j, k;
215	struct sysctl_oid **oidpp;
216
217	j = l->ls_length;
218	oidpp = (struct sysctl_oid **) l->ls_items;
219	for (; j--; oidpp++) {
220
221		if (!*oidpp)
222			continue;
223
224		for (k=0; k<i; k++)
225			printf(" ");
226
227		if ((*oidpp)->oid_number > 100) {
228			printf("Junk! %p  # %d  %s  k %x  a1 %p  a2 %x  h %p\n",
229				*oidpp,
230		 		(*oidpp)->oid_number, (*oidpp)->oid_name,
231		 		(*oidpp)->oid_kind, (*oidpp)->oid_arg1,
232		 		(*oidpp)->oid_arg2, (*oidpp)->oid_handler);
233			continue;
234		}
235		printf("%d %s ", (*oidpp)->oid_number, (*oidpp)->oid_name);
236
237		printf("%c%c",
238			(*oidpp)->oid_kind & CTLFLAG_RD ? 'R':' ',
239			(*oidpp)->oid_kind & CTLFLAG_WR ? 'W':' ');
240
241		switch ((*oidpp)->oid_kind & CTLTYPE) {
242			case CTLTYPE_NODE:
243				if ((*oidpp)->oid_handler) {
244					printf(" Node(proc)\n");
245				} else {
246					printf(" Node\n");
247					sysctl_sysctl_debug_dump_node(
248						(*oidpp)->oid_arg1, i+2);
249				}
250				break;
251			case CTLTYPE_INT:    printf(" Int\n"); break;
252			case CTLTYPE_STRING: printf(" String\n"); break;
253			case CTLTYPE_QUAD:   printf(" Quad\n"); break;
254			case CTLTYPE_OPAQUE: printf(" Opaque/struct\n"); break;
255			default:	     printf("\n");
256		}
257
258	}
259}
260
261
262static int
263sysctl_sysctl_debug SYSCTL_HANDLER_ARGS
264{
265	sysctl_sysctl_debug_dump_node(&sysctl_, 0);
266	return ENOENT;
267}
268
269SYSCTL_PROC(_sysctl, 0, debug, CTLTYPE_STRING|CTLFLAG_RD,
270	0, 0, sysctl_sysctl_debug, "");
271
272char domainname[MAXHOSTNAMELEN];
273int domainnamelen;
274static int
275sysctl_kern_domainname SYSCTL_HANDLER_ARGS
276{
277	int error = sysctl_handle_string(oidp,
278		oidp->oid_arg1, oidp->oid_arg2, req);
279	if (req->newptr && (error == 0 || error == ENOMEM))
280		domainnamelen = req->newlen;
281	return error;
282}
283
284SYSCTL_PROC(_kern, KERN_DOMAINNAME, domainname, CTLTYPE_STRING|CTLFLAG_RW,
285	&domainname, sizeof(domainname), sysctl_kern_domainname, "");
286
287long hostid;
288/* Some trouble here, if sizeof (int) != sizeof (long) */
289SYSCTL_INT(_kern, KERN_HOSTID, hostid, CTLFLAG_RW, &hostid, 0, "");
290
291/*
292 * Handle an integer, signed or unsigned.
293 * Two cases:
294 *     a variable:  point arg1 at it.
295 *     a constant:  pass it in arg2.
296 */
297
298int
299sysctl_handle_int SYSCTL_HANDLER_ARGS
300{
301	int error = 0;
302
303	if (arg1)
304		error = SYSCTL_OUT(req, arg1, sizeof(int));
305	else if (arg2)
306		error = SYSCTL_OUT(req, &arg2, sizeof(int));
307
308	if (error || !req->newptr)
309		return (error);
310
311	if (!arg1)
312		error = EPERM;
313	else
314		error = SYSCTL_IN(req, arg1, sizeof(int));
315	return (error);
316}
317
318/*
319 * Handle our generic '\0' terminated 'C' string.
320 * Two cases:
321 * 	a variable string:  point arg1 at it, arg2 is max length.
322 * 	a constant string:  point arg1 at it, arg2 is zero.
323 */
324
325int
326sysctl_handle_string SYSCTL_HANDLER_ARGS
327{
328	int error=0;
329
330	if (arg2)
331		error = SYSCTL_OUT(req, arg1, arg2);
332	else
333		error = SYSCTL_OUT(req, arg1, strlen((char *)arg1)+1);
334
335	if (error || !req->newptr || !arg2)
336		return (error);
337
338	if ((req->newlen - req->newidx) > arg2) {
339		error = E2BIG;
340	} else {
341		arg2 = (req->newlen - req->newidx);
342		error = SYSCTL_IN(req, arg1, arg2);
343		((char *)arg1)[arg2] = '\0';
344	}
345
346	return (error);
347}
348
349/*
350 * Handle any kind of opaque data.
351 * arg1 points to it, arg2 is the size.
352 */
353
354int
355sysctl_handle_opaque SYSCTL_HANDLER_ARGS
356{
357	int error;
358
359	error = SYSCTL_OUT(req, arg1, arg2);
360
361	if (error || !req->newptr)
362		return (error);
363
364	error = SYSCTL_IN(req, arg1, arg2);
365
366	return (error);
367}
368
369/*
370 * Transfer functions to/from kernel space.
371 * XXX: rather untested at this point
372 */
373static int
374sysctl_old_kernel(struct sysctl_req *req, void *p, int l)
375{
376	int i = 0;
377
378	if (req->oldptr) {
379		i = min(req->oldlen - req->oldidx, l);
380		if (i > 0)
381			bcopy(p, req->oldptr + req->oldidx, i);
382	}
383	req->oldidx += l;
384	if (i != l)
385		return (ENOMEM);
386	return (0);
387
388}
389
390static int
391sysctl_new_kernel(struct sysctl_req *req, void *p, int l)
392{
393	int i;
394	if (!req->newptr)
395		return 0;
396	if (req->newlen - req->newidx < l)
397		return (EINVAL);
398	bcopy(req->newptr + req->newidx, p, l);
399	req->newidx += l;
400	return (0);
401}
402
403/*
404 * Transfer function to/from user space.
405 */
406static int
407sysctl_old_user(struct sysctl_req *req, void *p, int l)
408{
409	int error = 0, i = 0;
410
411	if (req->oldptr) {
412		i = min(req->oldlen - req->oldidx, l);
413		if (i > 0)
414			error  = copyout(p, req->oldptr + req->oldidx, i);
415	}
416	req->oldidx += l;
417	if (error)
418		return (error);
419	if (req->oldptr && i < l)
420		return (ENOMEM);
421	return (0);
422}
423
424static int
425sysctl_new_user(struct sysctl_req *req, void *p, int l)
426{
427	int error, i;
428
429	if (!req->newptr)
430		return 0;
431	if (req->newlen - req->newidx < l)
432		return (EINVAL);
433	error = copyin(req->newptr + req->newidx, p, l);
434	req->newidx += l;
435	return (error);
436}
437
438#ifdef DEBUG
439static sysctlfn debug_sysctl;
440#endif
441
442/*
443 * Locking and stats
444 */
445static struct sysctl_lock {
446	int	sl_lock;
447	int	sl_want;
448	int	sl_locked;
449} memlock;
450
451
452
453/*
454 * Traverse our tree, and find the right node, execute whatever it points
455 * at, and return the resulting error code.
456 * We work entirely in kernel-space at this time.
457 */
458
459
460int
461sysctl_root SYSCTL_HANDLER_ARGS
462{
463	int *name = (int *) arg1;
464	int namelen = arg2;
465	int indx, i, j;
466	struct sysctl_oid **oidpp;
467	struct linker_set *lsp = &sysctl_;
468
469	j = lsp->ls_length;
470	oidpp = (struct sysctl_oid **) lsp->ls_items;
471
472	indx = 0;
473	while (j-- && indx < CTL_MAXNAME) {
474		if (*oidpp && ((*oidpp)->oid_number == name[indx])) {
475			indx++;
476			if (((*oidpp)->oid_kind & CTLTYPE) == CTLTYPE_NODE) {
477				if ((*oidpp)->oid_handler)
478					goto found;
479				if (indx == namelen)
480					return ENOENT;
481				lsp = (struct linker_set*)(*oidpp)->oid_arg1;
482				j = lsp->ls_length;
483				oidpp = (struct sysctl_oid **)lsp->ls_items;
484			} else {
485				if (indx != namelen)
486					return EISDIR;
487				goto found;
488			}
489		} else {
490			oidpp++;
491		}
492	}
493	return ENOENT;
494found:
495
496	/* If writing isn't allowed */
497	if (req->newptr && !((*oidpp)->oid_kind & CTLFLAG_WR))
498		return (EPERM);
499
500	if (!(*oidpp)->oid_handler)
501		return EINVAL;
502
503	if (((*oidpp)->oid_kind & CTLTYPE) == CTLTYPE_NODE) {
504		i = ((*oidpp)->oid_handler) (*oidpp,
505					name + indx, namelen - indx,
506					req);
507	} else {
508		i = ((*oidpp)->oid_handler) (*oidpp,
509					(*oidpp)->oid_arg1, (*oidpp)->oid_arg2,
510					req);
511	}
512	return (i);
513}
514
515#ifndef _SYS_SYSPROTO_H_
516struct sysctl_args {
517	int	*name;
518	u_int	namelen;
519	void	*old;
520	size_t	*oldlenp;
521	void	*new;
522	size_t	newlen;
523};
524#endif
525
526int
527__sysctl(p, uap, retval)
528	struct proc *p;
529	register struct sysctl_args *uap;
530	int *retval;
531{
532	int error, i, j, name[CTL_MAXNAME];
533
534	if (uap->namelen > CTL_MAXNAME || uap->namelen < 2)
535		return (EINVAL);
536
537 	error = copyin(uap->name, &name, uap->namelen * sizeof(int));
538 	if (error)
539		return (error);
540
541	error = userland_sysctl(p, name, uap->namelen,
542		uap->old, uap->oldlenp, 0,
543		uap->new, uap->newlen, &j);
544	if (error && error != ENOMEM)
545		return (error);
546	if (uap->oldlenp) {
547		i = copyout(&j, uap->oldlenp, sizeof(j));
548		if (i)
549			return (i);
550	}
551	return (error);
552}
553
554static sysctlfn kern_sysctl;
555
556/*
557 * This is used from various compatibility syscalls too.  That's why name
558 * must be in kernel space.
559 */
560int
561userland_sysctl(struct proc *p, int *name, u_int namelen, void *old, size_t *oldlenp, int inkernel, void *new, size_t newlen, int *retval)
562{
563	int error = 0, dolock = 1, i, oldlen = 0;
564	u_int savelen = 0;
565	sysctlfn *fn;
566	struct sysctl_req req;
567
568	bzero(&req, sizeof req);
569
570	if (new != NULL && (error = suser(p->p_ucred, &p->p_acflag)))
571		return (error);
572
573	if (oldlenp) {
574		if (inkernel) {
575			req.oldlen = *oldlenp;
576		} else {
577			error = copyin(oldlenp, &req.oldlen, sizeof(*oldlenp));
578			if (error)
579				return (error);
580		}
581	}
582
583	if (old) {
584		if (!useracc(old, req.oldlen, B_WRITE))
585			return (EFAULT);
586		req.oldptr= old;
587	}
588
589	if (newlen) {
590		if (!useracc(new, req.newlen, B_READ))
591			return (EFAULT);
592		req.newlen = newlen;
593		req.newptr = new;
594	}
595
596	req.oldfunc = sysctl_old_user;
597	req.newfunc = sysctl_new_user;
598
599	error = sysctl_root(0, name, namelen, &req);
600
601/*
602	if (error && error != ENOMEM)
603		return (error);
604*/
605	if (error == ENOENT)
606		goto oldstuff;
607
608	if (retval) {
609		if (req.oldptr && req.oldidx > req.oldlen)
610			*retval = req.oldlen;
611		else
612			*retval = req.oldidx;
613	}
614	return (error);
615
616oldstuff:
617	oldlen = req.oldlen;
618
619	switch (name[0]) {
620	case CTL_KERN:
621		fn = kern_sysctl;
622		if (name[1] != KERN_VNODE)      /* XXX */
623			dolock = 0;
624		break;
625	case CTL_HW:
626		fn = hw_sysctl;
627		break;
628	case CTL_VM:
629		fn = vm_sysctl;
630		break;
631	case CTL_NET:
632		fn = net_sysctl;
633		break;
634	case CTL_FS:
635		fn = fs_sysctl;
636		break;
637#ifdef DEBUG
638	case CTL_DEBUG:
639		fn = debug_sysctl;
640		break;
641#endif
642	default:
643		return (EOPNOTSUPP);
644	}
645	if (old != NULL) {
646		if (!useracc(old, oldlen, B_WRITE))
647			return (EFAULT);
648		while (memlock.sl_lock) {
649			memlock.sl_want = 1;
650			(void) tsleep((caddr_t)&memlock, PRIBIO+1, "sysctl", 0);
651			memlock.sl_locked++;
652		}
653		memlock.sl_lock = 1;
654		if (dolock)
655			vslock(old, oldlen);
656		savelen = oldlen;
657	}
658
659
660	error = (*fn)(name + 1, namelen - 1, old, &oldlen,
661	    new, newlen, p);
662
663
664	if (old != NULL) {
665		if (dolock)
666			vsunlock(old, savelen, B_WRITE);
667		memlock.sl_lock = 0;
668		if (memlock.sl_want) {
669			memlock.sl_want = 0;
670			wakeup((caddr_t)&memlock);
671		}
672	}
673#if 0
674	if (error) {
675		printf("SYSCTL_ERROR: ");
676		for(i=0;i<namelen;i++)
677			printf("%d ", name[i]);
678		printf("= %d\n", error);
679	}
680#endif
681	if (error)
682		return (error);
683	if (retval)
684		*retval = oldlen;
685	return (error);
686}
687
688/*
689 * Attributes stored in the kernel.
690 */
691int securelevel = -1;
692
693/*
694 * kernel related system variables.
695 */
696static int
697kern_sysctl(name, namelen, oldp, oldlenp, newp, newlen, p)
698	int *name;
699	u_int namelen;
700	void *oldp;
701	size_t *oldlenp;
702	void *newp;
703	size_t newlen;
704	struct proc *p;
705{
706	int error, level;
707	dev_t ndumpdev;
708
709	/* all sysctl names at this level are terminal */
710	if (namelen != 1 && !(name[0] == KERN_PROC || name[0] == KERN_PROF
711			      || name[0] == KERN_NTP_PLL))
712		return (ENOTDIR);		/* overloaded */
713
714	switch (name[0]) {
715
716	case KERN_SECURELVL:
717		level = securelevel;
718		if ((error = sysctl_int(oldp, oldlenp, newp, newlen, &level)) ||
719		    newp == NULL)
720			return (error);
721		if (level < securelevel && p->p_pid != 1)
722			return (EPERM);
723		securelevel = level;
724		return (0);
725	case KERN_VNODE:
726		return (sysctl_vnode(oldp, oldlenp));
727	case KERN_PROC:
728		return (sysctl_doproc(name + 1, namelen - 1, oldp, oldlenp));
729	case KERN_FILE:
730		return (sysctl_file(oldp, oldlenp));
731#ifdef GPROF
732	case KERN_PROF:
733		return (sysctl_doprof(name + 1, namelen - 1, oldp, oldlenp,
734		    newp, newlen));
735#endif
736	case KERN_NTP_PLL:
737		return (ntp_sysctl(name + 1, namelen - 1, oldp, oldlenp,
738				   newp, newlen, p));
739	case KERN_DUMPDEV:
740		ndumpdev = dumpdev;
741		error = sysctl_struct(oldp, oldlenp, newp, newlen, &ndumpdev,
742				      sizeof ndumpdev);
743		if (!error && ndumpdev != dumpdev) {
744			error = setdumpdev(ndumpdev);
745		}
746		return error;
747	default:
748		return (EOPNOTSUPP);
749	}
750	/* NOTREACHED */
751}
752
753/*
754 * hardware related system variables.
755 */
756int
757hw_sysctl(name, namelen, oldp, oldlenp, newp, newlen, p)
758	int *name;
759	u_int namelen;
760	void *oldp;
761	size_t *oldlenp;
762	void *newp;
763	size_t newlen;
764	struct proc *p;
765{
766	/* almost all sysctl names at this level are terminal */
767	if (namelen != 1 && name[0] != HW_DEVCONF)
768		return (ENOTDIR);		/* overloaded */
769
770	switch (name[0]) {
771	case HW_PHYSMEM:
772		return (sysctl_rdint(oldp, oldlenp, newp, ctob(physmem)));
773	case HW_USERMEM:
774		return (sysctl_rdint(oldp, oldlenp, newp,
775		    ctob(physmem - cnt.v_wire_count)));
776	case HW_DEVCONF:
777		return (dev_sysctl(name + 1, namelen - 1, oldp, oldlenp,
778				   newp, newlen, p));
779	default:
780		return (EOPNOTSUPP);
781	}
782	/* NOTREACHED */
783}
784
785#ifdef DEBUG
786/*
787 * Debugging related system variables.
788 */
789struct ctldebug debug0, debug1, debug2, debug3, debug4;
790struct ctldebug debug5, debug6, debug7, debug8, debug9;
791struct ctldebug debug10, debug11, debug12, debug13, debug14;
792struct ctldebug debug15, debug16, debug17, debug18, debug19;
793static struct ctldebug *debugvars[CTL_DEBUG_MAXID] = {
794	&debug0, &debug1, &debug2, &debug3, &debug4,
795	&debug5, &debug6, &debug7, &debug8, &debug9,
796	&debug10, &debug11, &debug12, &debug13, &debug14,
797	&debug15, &debug16, &debug17, &debug18, &debug19,
798};
799static int
800debug_sysctl(name, namelen, oldp, oldlenp, newp, newlen, p)
801	int *name;
802	u_int namelen;
803	void *oldp;
804	size_t *oldlenp;
805	void *newp;
806	size_t newlen;
807	struct proc *p;
808{
809	struct ctldebug *cdp;
810
811	/* all sysctl names at this level are name and field */
812	if (namelen != 2)
813		return (ENOTDIR);		/* overloaded */
814	cdp = debugvars[name[0]];
815	if (cdp->debugname == 0)
816		return (EOPNOTSUPP);
817	switch (name[1]) {
818	case CTL_DEBUG_NAME:
819		return (sysctl_rdstring(oldp, oldlenp, newp, cdp->debugname));
820	case CTL_DEBUG_VALUE:
821		return (sysctl_int(oldp, oldlenp, newp, newlen, cdp->debugvar));
822	default:
823		return (EOPNOTSUPP);
824	}
825	/* NOTREACHED */
826}
827#endif /* DEBUG */
828
829/*
830 * Validate parameters and get old / set new parameters
831 * for an integer-valued sysctl function.
832 */
833int
834sysctl_int(oldp, oldlenp, newp, newlen, valp)
835	void *oldp;
836	size_t *oldlenp;
837	void *newp;
838	size_t newlen;
839	int *valp;
840{
841	int error = 0;
842
843	if (oldp && *oldlenp < sizeof(int))
844		return (ENOMEM);
845	if (newp && newlen != sizeof(int))
846		return (EINVAL);
847	*oldlenp = sizeof(int);
848	if (oldp)
849		error = copyout(valp, oldp, sizeof(int));
850	if (error == 0 && newp)
851		error = copyin(newp, valp, sizeof(int));
852	return (error);
853}
854
855/*
856 * As above, but read-only.
857 */
858int
859sysctl_rdint(oldp, oldlenp, newp, val)
860	void *oldp;
861	size_t *oldlenp;
862	void *newp;
863	int val;
864{
865	int error = 0;
866
867	if (oldp && *oldlenp < sizeof(int))
868		return (ENOMEM);
869	if (newp)
870		return (EPERM);
871	*oldlenp = sizeof(int);
872	if (oldp)
873		error = copyout((caddr_t)&val, oldp, sizeof(int));
874	return (error);
875}
876
877/*
878 * Validate parameters and get old / set new parameters
879 * for a string-valued sysctl function.
880 */
881int
882sysctl_string(oldp, oldlenp, newp, newlen, str, maxlen)
883	void *oldp;
884	size_t *oldlenp;
885	void *newp;
886	size_t newlen;
887	char *str;
888	int maxlen;
889{
890	int len, error = 0, rval = 0;
891
892	len = strlen(str) + 1;
893	if (oldp && *oldlenp < len) {
894		len = *oldlenp;
895		rval = ENOMEM;
896	}
897	if (newp && newlen >= maxlen)
898		return (EINVAL);
899	if (oldp) {
900		*oldlenp = len;
901		error = copyout(str, oldp, len);
902		if (error)
903			rval = error;
904	}
905	if ((error == 0 || error == ENOMEM) && newp) {
906		error = copyin(newp, str, newlen);
907		if (error)
908			rval = error;
909		str[newlen] = 0;
910	}
911	return (rval);
912}
913
914/*
915 * As above, but read-only.
916 */
917int
918sysctl_rdstring(oldp, oldlenp, newp, str)
919	void *oldp;
920	size_t *oldlenp;
921	void *newp;
922	char *str;
923{
924	int len, error = 0, rval = 0;
925
926	len = strlen(str) + 1;
927	if (oldp && *oldlenp < len) {
928		len = *oldlenp;
929		rval = ENOMEM;
930	}
931	if (newp)
932		return (EPERM);
933	*oldlenp = len;
934	if (oldp)
935		error = copyout(str, oldp, len);
936		if (error)
937			rval = error;
938	return (rval);
939}
940
941/*
942 * Validate parameters and get old / set new parameters
943 * for a structure oriented sysctl function.
944 */
945int
946sysctl_struct(oldp, oldlenp, newp, newlen, sp, len)
947	void *oldp;
948	size_t *oldlenp;
949	void *newp;
950	size_t newlen;
951	void *sp;
952	int len;
953{
954	int error = 0;
955
956	if (oldp && *oldlenp < len)
957		return (ENOMEM);
958	if (newp && newlen > len)
959		return (EINVAL);
960	if (oldp) {
961		*oldlenp = len;
962		error = copyout(sp, oldp, len);
963	}
964	if (error == 0 && newp)
965		error = copyin(newp, sp, len);
966	return (error);
967}
968
969/*
970 * Validate parameters and get old parameters
971 * for a structure oriented sysctl function.
972 */
973int
974sysctl_rdstruct(oldp, oldlenp, newp, sp, len)
975	void *oldp;
976	size_t *oldlenp;
977	void *newp, *sp;
978	int len;
979{
980	int error = 0;
981
982	if (oldp && *oldlenp < len)
983		return (ENOMEM);
984	if (newp)
985		return (EPERM);
986	*oldlenp = len;
987	if (oldp)
988		error = copyout(sp, oldp, len);
989	return (error);
990}
991
992/*
993 * Get file structures.
994 */
995int
996sysctl_file(where, sizep)
997	char *where;
998	size_t *sizep;
999{
1000	int buflen, error;
1001	struct file *fp;
1002	char *start = where;
1003
1004	buflen = *sizep;
1005	if (where == NULL) {
1006		/*
1007		 * overestimate by 10 files
1008		 */
1009		*sizep = sizeof(filehead) + (nfiles + 10) * sizeof(struct file);
1010		return (0);
1011	}
1012
1013	/*
1014	 * first copyout filehead
1015	 */
1016	if (buflen < sizeof(filehead)) {
1017		*sizep = 0;
1018		return (0);
1019	}
1020	error = copyout((caddr_t)&filehead, where, sizeof(filehead));
1021	if (error)
1022		return (error);
1023	buflen -= sizeof(filehead);
1024	where += sizeof(filehead);
1025
1026	/*
1027	 * followed by an array of file structures
1028	 */
1029	for (fp = filehead; fp != NULL; fp = fp->f_filef) {
1030		if (buflen < sizeof(struct file)) {
1031			*sizep = where - start;
1032			return (ENOMEM);
1033		}
1034		error = copyout((caddr_t)fp, where, sizeof (struct file));
1035		if (error)
1036			return (error);
1037		buflen -= sizeof(struct file);
1038		where += sizeof(struct file);
1039	}
1040	*sizep = where - start;
1041	return (0);
1042}
1043
1044/*
1045 * try over estimating by 5 procs
1046 */
1047#define KERN_PROCSLOP	(5 * sizeof (struct kinfo_proc))
1048
1049int
1050sysctl_doproc(name, namelen, where, sizep)
1051	int *name;
1052	u_int namelen;
1053	char *where;
1054	size_t *sizep;
1055{
1056	register struct proc *p;
1057	register struct kinfo_proc *dp = (struct kinfo_proc *)where;
1058	register int needed = 0;
1059	int buflen = where != NULL ? *sizep : 0;
1060	int doingzomb;
1061	struct eproc eproc;
1062	int error = 0;
1063
1064	if (namelen != 2 && !(namelen == 1 && name[0] == KERN_PROC_ALL))
1065		return (EINVAL);
1066	p = (struct proc *)allproc;
1067	doingzomb = 0;
1068again:
1069	for (; p != NULL; p = p->p_next) {
1070		/*
1071		 * Skip embryonic processes.
1072		 */
1073		if (p->p_stat == SIDL)
1074			continue;
1075		/*
1076		 * TODO - make more efficient (see notes below).
1077		 * do by session.
1078		 */
1079		switch (name[0]) {
1080
1081		case KERN_PROC_PID:
1082			/* could do this with just a lookup */
1083			if (p->p_pid != (pid_t)name[1])
1084				continue;
1085			break;
1086
1087		case KERN_PROC_PGRP:
1088			/* could do this by traversing pgrp */
1089			if (p->p_pgrp == NULL || p->p_pgrp->pg_id != (pid_t)name[1])
1090				continue;
1091			break;
1092
1093		case KERN_PROC_TTY:
1094			if ((p->p_flag & P_CONTROLT) == 0 ||
1095			    p->p_session == NULL ||
1096			    p->p_session->s_ttyp == NULL ||
1097			    p->p_session->s_ttyp->t_dev != (dev_t)name[1])
1098				continue;
1099			break;
1100
1101		case KERN_PROC_UID:
1102			if (p->p_ucred == NULL || p->p_ucred->cr_uid != (uid_t)name[1])
1103				continue;
1104			break;
1105
1106		case KERN_PROC_RUID:
1107			if (p->p_ucred == NULL || p->p_cred->p_ruid != (uid_t)name[1])
1108				continue;
1109			break;
1110		}
1111		if (buflen >= sizeof(struct kinfo_proc)) {
1112			fill_eproc(p, &eproc);
1113			error = copyout((caddr_t)p, &dp->kp_proc,
1114			    sizeof(struct proc));
1115			if (error)
1116				return (error);
1117			error = copyout((caddr_t)&eproc, &dp->kp_eproc,
1118			    sizeof(eproc));
1119			if (error)
1120				return (error);
1121			dp++;
1122			buflen -= sizeof(struct kinfo_proc);
1123		}
1124		needed += sizeof(struct kinfo_proc);
1125	}
1126	if (doingzomb == 0) {
1127		p = zombproc;
1128		doingzomb++;
1129		goto again;
1130	}
1131	if (where != NULL) {
1132		*sizep = (caddr_t)dp - where;
1133		if (needed > *sizep)
1134			return (ENOMEM);
1135	} else {
1136		needed += KERN_PROCSLOP;
1137		*sizep = needed;
1138	}
1139	return (0);
1140}
1141
1142/*
1143 * Fill in an eproc structure for the specified process.
1144 */
1145void
1146fill_eproc(p, ep)
1147	register struct proc *p;
1148	register struct eproc *ep;
1149{
1150	register struct tty *tp;
1151
1152	bzero(ep, sizeof(*ep));
1153
1154	ep->e_paddr = p;
1155	if (p->p_cred) {
1156		ep->e_pcred = *p->p_cred;
1157		if (p->p_ucred)
1158			ep->e_ucred = *p->p_ucred;
1159	}
1160	if (p->p_stat != SIDL && p->p_stat != SZOMB && p->p_vmspace != NULL) {
1161		register struct vmspace *vm = p->p_vmspace;
1162
1163#ifdef pmap_resident_count
1164		ep->e_vm.vm_rssize = pmap_resident_count(&vm->vm_pmap); /*XXX*/
1165#else
1166		ep->e_vm.vm_rssize = vm->vm_rssize;
1167#endif
1168		ep->e_vm.vm_tsize = vm->vm_tsize;
1169		ep->e_vm.vm_dsize = vm->vm_dsize;
1170		ep->e_vm.vm_ssize = vm->vm_ssize;
1171#ifndef sparc
1172		ep->e_vm.vm_pmap = vm->vm_pmap;
1173#endif
1174	}
1175	if (p->p_pptr)
1176		ep->e_ppid = p->p_pptr->p_pid;
1177	if (p->p_pgrp) {
1178		ep->e_sess = p->p_pgrp->pg_session;
1179		ep->e_pgid = p->p_pgrp->pg_id;
1180		ep->e_jobc = p->p_pgrp->pg_jobc;
1181	}
1182	if ((p->p_flag & P_CONTROLT) &&
1183	    (ep->e_sess != NULL) &&
1184	    ((tp = ep->e_sess->s_ttyp) != NULL)) {
1185		ep->e_tdev = tp->t_dev;
1186		ep->e_tpgid = tp->t_pgrp ? tp->t_pgrp->pg_id : NO_PID;
1187		ep->e_tsess = tp->t_session;
1188	} else
1189		ep->e_tdev = NODEV;
1190	if (ep->e_sess && ep->e_sess->s_ttyvp)
1191		ep->e_flag = EPROC_CTTY;
1192	if (SESS_LEADER(p))
1193		ep->e_flag |= EPROC_SLEADER;
1194	if (p->p_wmesg) {
1195		strncpy(ep->e_wmesg, p->p_wmesg, WMESGLEN);
1196		ep->e_wmesg[WMESGLEN] = 0;
1197	}
1198}
1199
1200#ifdef COMPAT_43
1201#include <sys/socket.h>
1202#define	KINFO_PROC		(0<<8)
1203#define	KINFO_RT		(1<<8)
1204#define	KINFO_VNODE		(2<<8)
1205#define	KINFO_FILE		(3<<8)
1206#define	KINFO_METER		(4<<8)
1207#define	KINFO_LOADAVG		(5<<8)
1208#define	KINFO_CLOCKRATE		(6<<8)
1209
1210/* Non-standard BSDI extension - only present on their 4.3 net-2 releases */
1211#define	KINFO_BSDI_SYSINFO	(101<<8)
1212
1213/*
1214 * XXX this is bloat, but I hope it's better here than on the potentially
1215 * limited kernel stack...  -Peter
1216 */
1217
1218struct {
1219	int	bsdi_machine;		/* "i386" on BSD/386 */
1220/*      ^^^ this is an offset to the string, relative to the struct start */
1221	char	*pad0;
1222	long	pad1;
1223	long	pad2;
1224	long	pad3;
1225	u_long	pad4;
1226	u_long	pad5;
1227	u_long	pad6;
1228
1229	int	bsdi_ostype;		/* "BSD/386" on BSD/386 */
1230	int	bsdi_osrelease;		/* "1.1" on BSD/386 */
1231	long	pad7;
1232	long	pad8;
1233	char	*pad9;
1234
1235	long	pad10;
1236	long	pad11;
1237	int	pad12;
1238	long	pad13;
1239	quad_t	pad14;
1240	long	pad15;
1241
1242	struct	timeval pad16;
1243	/* we dont set this, because BSDI's uname used gethostname() instead */
1244	int	bsdi_hostname;		/* hostname on BSD/386 */
1245
1246	/* the actual string data is appended here */
1247
1248} bsdi_si;
1249/*
1250 * this data is appended to the end of the bsdi_si structure during copyout.
1251 * The "char *" offsets are relative to the base of the bsdi_si struct.
1252 * This contains "FreeBSD\02.0-BUILT-nnnnnn\0i386\0", and these strings
1253 * should not exceed the length of the buffer here... (or else!! :-)
1254 */
1255char bsdi_strings[80];	/* It had better be less than this! */
1256
1257#ifndef _SYS_SYSPROTO_H_
1258struct getkerninfo_args {
1259	int	op;
1260	char	*where;
1261	int	*size;
1262	int	arg;
1263};
1264#endif
1265
1266int
1267ogetkerninfo(p, uap, retval)
1268	struct proc *p;
1269	register struct getkerninfo_args *uap;
1270	int *retval;
1271{
1272	int error, name[6];
1273	u_int size;
1274
1275	switch (uap->op & 0xff00) {
1276
1277	case KINFO_RT:
1278		name[0] = CTL_NET;
1279		name[1] = PF_ROUTE;
1280		name[2] = 0;
1281		name[3] = (uap->op & 0xff0000) >> 16;
1282		name[4] = uap->op & 0xff;
1283		name[5] = uap->arg;
1284		error = userland_sysctl(p, name, 6, uap->where, uap->size,
1285			0, 0, 0, 0);
1286		break;
1287
1288	case KINFO_VNODE:
1289		name[0] = CTL_KERN;
1290		name[1] = KERN_VNODE;
1291		error = userland_sysctl(p, name, 2, uap->where, uap->size,
1292			0, 0, 0, 0);
1293		break;
1294
1295	case KINFO_PROC:
1296		name[0] = CTL_KERN;
1297		name[1] = KERN_PROC;
1298		name[2] = uap->op & 0xff;
1299		name[3] = uap->arg;
1300		error = userland_sysctl(p, name, 4, uap->where, uap->size,
1301			0, 0, 0, 0);
1302		break;
1303
1304	case KINFO_FILE:
1305		name[0] = CTL_KERN;
1306		name[1] = KERN_FILE;
1307		error = userland_sysctl(p, name, 2, uap->where, uap->size,
1308			0, 0, 0, 0);
1309		break;
1310
1311	case KINFO_METER:
1312		name[0] = CTL_VM;
1313		name[1] = VM_METER;
1314		error = userland_sysctl(p, name, 2, uap->where, uap->size,
1315			0, 0, 0, 0);
1316		break;
1317
1318	case KINFO_LOADAVG:
1319		name[0] = CTL_VM;
1320		name[1] = VM_LOADAVG;
1321		error = userland_sysctl(p, name, 2, uap->where, uap->size,
1322			0, 0, 0, 0);
1323		break;
1324
1325	case KINFO_CLOCKRATE:
1326		name[0] = CTL_KERN;
1327		name[1] = KERN_CLOCKRATE;
1328		error = userland_sysctl(p, name, 2, uap->where, uap->size,
1329			0, 0, 0, 0);
1330		break;
1331
1332	case KINFO_BSDI_SYSINFO: {
1333		/*
1334		 * this is pretty crude, but it's just enough for uname()
1335		 * from BSDI's 1.x libc to work.
1336		 *
1337		 * In particular, it doesn't return the same results when
1338		 * the supplied buffer is too small.  BSDI's version apparently
1339		 * will return the amount copied, and set the *size to how
1340		 * much was needed.  The emulation framework here isn't capable
1341		 * of that, so we just set both to the amount copied.
1342		 * BSDI's 2.x product apparently fails with ENOMEM in this
1343		 * scenario.
1344		 */
1345
1346		u_int needed;
1347		u_int left;
1348		char *s;
1349
1350		bzero((char *)&bsdi_si, sizeof(bsdi_si));
1351		bzero(bsdi_strings, sizeof(bsdi_strings));
1352
1353		s = bsdi_strings;
1354
1355		bsdi_si.bsdi_ostype = (s - bsdi_strings) + sizeof(bsdi_si);
1356		strcpy(s, ostype);
1357		s += strlen(s) + 1;
1358
1359		bsdi_si.bsdi_osrelease = (s - bsdi_strings) + sizeof(bsdi_si);
1360		strcpy(s, osrelease);
1361		s += strlen(s) + 1;
1362
1363		bsdi_si.bsdi_machine = (s - bsdi_strings) + sizeof(bsdi_si);
1364		strcpy(s, machine);
1365		s += strlen(s) + 1;
1366
1367		needed = sizeof(bsdi_si) + (s - bsdi_strings);
1368
1369		if (uap->where == NULL) {
1370			/* process is asking how much buffer to supply.. */
1371			size = needed;
1372			error = 0;
1373			break;
1374		}
1375
1376
1377		/* if too much buffer supplied, trim it down */
1378		if (size > needed)
1379			size = needed;
1380
1381		/* how much of the buffer is remaining */
1382		left = size;
1383
1384		if ((error = copyout((char *)&bsdi_si, uap->where, left)) != 0)
1385			break;
1386
1387		/* is there any point in continuing? */
1388		if (left > sizeof(bsdi_si)) {
1389			left -= sizeof(bsdi_si);
1390			error = copyout(&bsdi_strings,
1391					uap->where + sizeof(bsdi_si), left);
1392		}
1393		break;
1394	}
1395
1396	default:
1397		return (EOPNOTSUPP);
1398	}
1399	if (error)
1400		return (error);
1401	*retval = size;
1402	if (uap->size)
1403		error = copyout((caddr_t)&size, (caddr_t)uap->size,
1404		    sizeof(size));
1405	return (error);
1406}
1407#endif /* COMPAT_43 */
1408