kern_sysctl.c revision 12187
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 1.33 1995/11/09 20:22:09 phk Exp $
38 */
39
40/*
41 * sysctl system call.
42 */
43
44#include <sys/param.h>
45#include <sys/systm.h>
46#include <sys/kernel.h>
47#include <sys/malloc.h>
48#include <sys/proc.h>
49#include <sys/file.h>
50#include <sys/vnode.h>
51#include <sys/unistd.h>
52#include <sys/buf.h>
53#include <sys/ioctl.h>
54#include <sys/tty.h>
55#include <sys/conf.h>
56#include <vm/vm.h>
57#include <sys/sysctl.h>
58#include <sys/user.h>
59
60extern struct linker_set sysctl_;
61
62/* BEGIN_MIB */
63SYSCTL_NODE(, 0,	  sysctl, CTLFLAG_RW, 0,
64	"Sysctl internal magic");
65SYSCTL_NODE(, CTL_KERN,	  kern,   CTLFLAG_RW, 0,
66	"High kernel, proc, limits &c");
67SYSCTL_NODE(, CTL_VM,	  vm,     CTLFLAG_RW, 0,
68	"Virtual memory");
69SYSCTL_NODE(, CTL_FS,	  fs,     CTLFLAG_RW, 0,
70	"File system");
71SYSCTL_NODE(, CTL_NET,	  net,    CTLFLAG_RW, 0,
72	"Network, (see socket.h)");
73SYSCTL_NODE(, CTL_DEBUG,  debug,  CTLFLAG_RW, 0,
74	"Debugging");
75SYSCTL_NODE(, CTL_HW,	  hw,     CTLFLAG_RW, 0,
76	"hardware");
77SYSCTL_NODE(, CTL_MACHDEP, machdep, CTLFLAG_RW, 0,
78	"machine dependent");
79SYSCTL_NODE(, CTL_USER,	  user,   CTLFLAG_RW, 0,
80	"user-level");
81
82SYSCTL_STRING(_kern, KERN_OSRELEASE, osrelease, CTLFLAG_RD, osrelease, 0, "");
83
84SYSCTL_INT(_kern, KERN_OSREV, osrevision, CTLFLAG_RD, 0, BSD, "");
85
86SYSCTL_STRING(_kern, KERN_VERSION, version, CTLFLAG_RD, version, 0, "");
87
88SYSCTL_STRING(_kern, KERN_OSTYPE, ostype, CTLFLAG_RD, ostype, 0, "");
89
90extern int osreldate;
91SYSCTL_INT(_kern, KERN_OSRELDATE, osreldate, CTLFLAG_RD, &osreldate, 0, "");
92
93SYSCTL_INT(_kern, KERN_MAXVNODES, maxvnodes, CTLFLAG_RD, &desiredvnodes, 0, "");
94
95SYSCTL_INT(_kern, KERN_MAXPROC, maxproc, CTLFLAG_RD, &maxproc, 0, "");
96
97SYSCTL_INT(_kern, KERN_MAXPROCPERUID, maxprocperuid,
98	CTLFLAG_RD, &maxprocperuid, 0, "");
99
100SYSCTL_INT(_kern, KERN_MAXFILESPERPROC, maxfilesperproc,
101	CTLFLAG_RD, &maxfilesperproc, 0, "");
102
103SYSCTL_INT(_kern, KERN_ARGMAX, argmax, CTLFLAG_RD, 0, ARG_MAX, "");
104
105SYSCTL_INT(_kern, KERN_POSIX1, posix1version, CTLFLAG_RD, 0, _POSIX_VERSION, "");
106
107SYSCTL_INT(_kern, KERN_NGROUPS, ngroups, CTLFLAG_RD, 0, NGROUPS_MAX, "");
108
109SYSCTL_INT(_kern, KERN_JOB_CONTROL, job_control, CTLFLAG_RD, 0, 1, "");
110
111SYSCTL_INT(_kern, KERN_MAXFILES, maxfiles, CTLFLAG_RW, &maxfiles, 0, "");
112
113#ifdef _POSIX_SAVED_IDS
114SYSCTL_INT(_kern, KERN_SAVED_IDS, saved_ids, CTLFLAG_RD, 0, 1, "");
115#else
116SYSCTL_INT(_kern, KERN_SAVED_IDS, saved_ids, CTLFLAG_RD, 0, 0, "");
117#endif
118
119char kernelname[MAXPATHLEN] = "/kernel";	/* XXX bloat */
120
121SYSCTL_STRING(_kern, KERN_BOOTFILE, bootfile,
122	CTLFLAG_RW, kernelname, sizeof kernelname, "");
123
124SYSCTL_STRUCT(_kern, KERN_BOOTTIME, boottime,
125	CTLFLAG_RW, &boottime, timeval, "");
126
127SYSCTL_STRING(_hw, HW_MACHINE, machine, CTLFLAG_RD, machine, 0, "");
128
129SYSCTL_STRING(_hw, HW_MODEL, model, CTLFLAG_RD, cpu_model, 0, "");
130
131SYSCTL_INT(_hw, HW_NCPU, ncpu, CTLFLAG_RD, 0, 1, "");
132
133SYSCTL_INT(_hw, HW_BYTEORDER, byteorder, CTLFLAG_RD, 0, BYTE_ORDER, "");
134
135SYSCTL_INT(_hw, HW_PAGESIZE, pagesize, CTLFLAG_RD, 0, PAGE_SIZE, "");
136
137/* END_MIB */
138
139extern int vfs_update_wakeup;
140extern int vfs_update_interval;
141static int
142sysctl_kern_updateinterval SYSCTL_HANDLER_ARGS
143{
144	int error = sysctl_handle_int(oidp,
145		oidp->oid_arg1, oidp->oid_arg2,
146		oldp, oldlenp, newp, newlen);
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,
163		oldp, oldlenp, newp, newlen);
164	if (newp && (error == 0 || error == ENOMEM))
165		hostnamelen = newlen;
166	return error;
167}
168
169SYSCTL_PROC(_kern, KERN_HOSTNAME, hostname, CTLTYPE_STRING|CTLFLAG_RW,
170	&hostname, sizeof(hostname), sysctl_kern_hostname, "");
171
172static int
173sysctl_order_cmp(void *a, void *b)
174{
175	struct sysctl_oid **pa,**pb;
176	pa = (struct sysctl_oid**) a;
177	pb = (struct sysctl_oid**) b;
178	if (!*pa) return 1;
179	if (!*pb) return -1;
180	return ((*pa)->oid_number - (*pb)->oid_number);
181}
182
183static void
184sysctl_order(void *arg)
185{
186	int j;
187	struct linker_set *l = (struct linker_set *) arg;
188	struct sysctl_oid **oidpp;
189
190	j = l->ls_length;
191	oidpp = (struct sysctl_oid **) l->ls_items;
192	for (; j--; oidpp++) {
193		if (!*oidpp)
194			continue;
195		if ((*oidpp)->oid_arg1 == arg) {
196			*oidpp = 0;
197			continue;
198		}
199		if (((*oidpp)->oid_kind & CTLTYPE) == CTLTYPE_NODE)
200			if (!(*oidpp)->oid_handler)
201				sysctl_order((*oidpp)->oid_arg1);
202	}
203	qsort(l->ls_items, l->ls_length, sizeof l->ls_items[0],
204		sysctl_order_cmp);
205}
206
207SYSINIT(sysctl,SI_SUB_KMEM,SI_ORDER_ANY,sysctl_order,&sysctl_);
208
209static void
210sysctl_sysctl_debug_dump_node(struct linker_set *l,int i)
211{
212	int j,k;
213	struct sysctl_oid **oidpp;
214
215	j = l->ls_length;
216	oidpp = (struct sysctl_oid **) l->ls_items;
217	for (; j--; oidpp++) {
218
219		if (!*oidpp)
220			continue;
221
222		for (k=0; k<i; k++)
223			printf(" ");
224
225		if ((*oidpp)->oid_number > 100) {
226			printf("Junk! %p  # %d  %s  k %x  a1 %p  a2 %x  h %p\n",
227				*oidpp,
228		 		(*oidpp)->oid_number, (*oidpp)->oid_name,
229		 		(*oidpp)->oid_kind, (*oidpp)->oid_arg1,
230		 		(*oidpp)->oid_arg2, (*oidpp)->oid_handler);
231			continue;
232		}
233		printf("%d %s ", (*oidpp)->oid_number, (*oidpp)->oid_name);
234
235		printf("%c%c",
236			(*oidpp)->oid_kind & CTLFLAG_RD ? 'R':' ',
237			(*oidpp)->oid_kind & CTLFLAG_WR ? 'W':' ');
238
239		switch ((*oidpp)->oid_kind & CTLTYPE) {
240			case CTLTYPE_NODE:
241				if ((*oidpp)->oid_handler) {
242					printf(" Node(proc)\n");
243				} else {
244					printf(" Node\n");
245					sysctl_sysctl_debug_dump_node(
246						(*oidpp)->oid_arg1,i+2);
247				}
248				break;
249			case CTLTYPE_INT:    printf(" Int\n"); break;
250			case CTLTYPE_STRING: printf(" String\n"); break;
251			case CTLTYPE_QUAD:   printf(" Quad\n"); break;
252			case CTLTYPE_OPAQUE: printf(" Opaque/struct\n"); break;
253			default:	     printf("\n");
254		}
255
256	}
257}
258
259
260static int
261sysctl_sysctl_debug SYSCTL_HANDLER_ARGS
262{
263	sysctl_sysctl_debug_dump_node(&sysctl_,0);
264	return ENOENT;
265}
266
267SYSCTL_PROC(_sysctl, 0, debug, CTLTYPE_STRING|CTLFLAG_RD,
268	0, 0, sysctl_sysctl_debug, "");
269
270char domainname[MAXHOSTNAMELEN];
271int domainnamelen;
272static int
273sysctl_kern_domainname SYSCTL_HANDLER_ARGS
274{
275	int error = sysctl_handle_string(oidp,
276		oidp->oid_arg1, oidp->oid_arg2,
277		oldp, oldlenp, newp, newlen);
278	if (newp && (error == 0 || error == ENOMEM))
279		domainnamelen = newlen;
280	return error;
281}
282
283SYSCTL_PROC(_kern, KERN_DOMAINNAME, domainname, CTLTYPE_STRING|CTLFLAG_RW,
284	&domainname, sizeof(domainname), sysctl_kern_domainname, "");
285
286long hostid;
287/* Some trouble here, if sizeof (int) != sizeof (long) */
288SYSCTL_INT(_kern, KERN_HOSTID, hostid, CTLFLAG_RW, &hostid, 0, "");
289
290int
291sysctl_handle_int SYSCTL_HANDLER_ARGS
292{
293	/* If there isn't sufficient space to return */
294	if (oldp && *oldlenp < sizeof(int))
295		return (ENOMEM);
296
297	/* If it is a constant, don't write */
298	if (newp && !arg1)
299		return (EPERM);
300
301	/* If we get more than an int */
302	if (newp && newlen != sizeof(int))
303		return (EINVAL);
304
305	*oldlenp = sizeof(int);
306	if (oldp && arg1 )
307		bcopy(arg1, oldp, sizeof(int));
308	else if (oldp)
309		bcopy(&arg2, oldp, sizeof(int));
310	if (newp)
311		bcopy(newp, arg1, sizeof(int));
312	return (0);
313}
314
315int
316sysctl_handle_string SYSCTL_HANDLER_ARGS
317{
318	int len, error=0;
319	char *str = (char *)arg1;
320
321	len = strlen(str) + 1;
322
323	if (oldp && *oldlenp < len) {
324		len = *oldlenp;
325		error=ENOMEM;
326	}
327
328	if (newp && newlen >= arg2)
329		return (EINVAL);
330
331	if (oldp) {
332		*oldlenp = len;
333		bcopy(str, oldp, len);
334	}
335
336	if (newp) {
337		bcopy(newp, str, newlen);
338		str[newlen] = 0;
339	}
340	return (error);
341}
342
343int
344sysctl_handle_opaque SYSCTL_HANDLER_ARGS
345{
346	if (oldp && *oldlenp < arg2)
347		return (ENOMEM);
348
349	if (newp && newlen != arg2)
350		return (EINVAL);
351
352	if (oldp) {
353		*oldlenp = arg2;
354		bcopy(arg1, oldp, arg2);
355	}
356	if (newp)
357		bcopy(newp, arg1, arg2);
358	return (0);
359}
360
361#ifdef DEBUG
362static sysctlfn debug_sysctl;
363#endif
364
365/*
366 * Locking and stats
367 */
368static struct sysctl_lock {
369	int	sl_lock;
370	int	sl_want;
371	int	sl_locked;
372} memlock;
373
374
375
376/*
377 * Traverse our tree, and find the right node, execute whatever it points
378 * at, and return the resulting error code.
379 * We work entirely in kernel-space at this time.
380 */
381
382
383int
384sysctl_root SYSCTL_HANDLER_ARGS
385{
386	int *name = (int *) arg1;
387	int namelen = arg2;
388	int indx, i, j;
389	struct sysctl_oid **oidpp;
390	struct linker_set *lsp = &sysctl_;
391
392	j = lsp->ls_length;
393	oidpp = (struct sysctl_oid **) lsp->ls_items;
394
395	indx = 0;
396	while (j-- && indx < CTL_MAXNAME) {
397		if (*oidpp && ((*oidpp)->oid_number == name[indx])) {
398			indx++;
399			if (((*oidpp)->oid_kind & CTLTYPE) == CTLTYPE_NODE) {
400				if ((*oidpp)->oid_handler)
401					goto found;
402				if (indx == namelen)
403					return ENOENT;
404				lsp = (struct linker_set*)(*oidpp)->oid_arg1;
405				j = lsp->ls_length;
406				oidpp = (struct sysctl_oid **)lsp->ls_items;
407			} else {
408				if (indx != namelen)
409					return EISDIR;
410				goto found;
411			}
412		} else {
413			oidpp++;
414		}
415	}
416	return EJUSTRETURN;
417found:
418
419	/* If writing isn't allowed */
420	if (newp && !((*oidpp)->oid_kind & CTLFLAG_WR))
421		return (EPERM);
422
423	if (!(*oidpp)->oid_handler)
424		return EINVAL;
425
426	if (((*oidpp)->oid_kind & CTLTYPE) == CTLTYPE_NODE) {
427		i = ((*oidpp)->oid_handler) (*oidpp,
428					name + indx, namelen - indx,
429					oldp, oldlenp, newp, newlen);
430	} else {
431		i = ((*oidpp)->oid_handler) (*oidpp,
432					(*oidpp)->oid_arg1, (*oidpp)->oid_arg2,
433					oldp, oldlenp, newp, newlen);
434	}
435	return (i);
436}
437
438struct sysctl_args {
439	int	*name;
440	u_int	namelen;
441	void	*old;
442	size_t	*oldlenp;
443	void	*new;
444	size_t	newlen;
445};
446
447int
448__sysctl(p, uap, retval)
449	struct proc *p;
450	register struct sysctl_args *uap;
451	int *retval;
452{
453	int error, name[CTL_MAXNAME];
454
455	if (uap->namelen > CTL_MAXNAME || uap->namelen < 2)
456		return (EINVAL);
457
458 	error = copyin(uap->name, &name, uap->namelen * sizeof(int));
459 	if (error)
460		return (error);
461
462	return (userland_sysctl(p, name, uap->namelen,
463		uap->old, uap->oldlenp, 0,
464		uap->new, uap->newlen, retval));
465}
466
467static sysctlfn kern_sysctl;
468
469/*
470 * This is used from various compatibility syscalls too.  That's why name
471 * must be in kernel space.
472 */
473int
474userland_sysctl(struct proc *p, int *name, u_int namelen, void *old, size_t *oldlenp, int inkernel, void *new, size_t newlen, int *retval)
475{
476	int error = 0, dolock = 1, i;
477	u_int savelen = 0, oldlen = 0;
478	sysctlfn *fn;
479	void *oldp = 0;
480	void *newp = 0;
481
482	if (new != NULL && (error = suser(p->p_ucred, &p->p_acflag)))
483		return (error);
484
485	if (oldlenp) {
486		if (inkernel) {
487			oldlen = *oldlenp;
488		} else {
489			error = copyin(oldlenp, &oldlen, sizeof(oldlen));
490			if (error)
491				return (error);
492		}
493	}
494
495	if (old)
496		oldp = malloc(oldlen, M_TEMP, M_WAITOK);
497
498	if (newlen) {
499		newp = malloc(newlen, M_TEMP, M_WAITOK);
500		error = copyin(new, newp, newlen);
501	}
502	if (error) {
503		if (oldp)
504			free(oldp, M_TEMP);
505		if (newp)
506			free(newp, M_TEMP);
507		return error;
508	}
509
510	error = sysctl_root(0, name, namelen, oldp, &oldlen,
511	newp, newlen);
512
513        if (!error || error == ENOMEM) {
514		if (oldlenp) {
515			if (inkernel) {
516				*oldlenp = oldlen;
517			} else {
518				i = copyout(&oldlen, oldlenp, sizeof(oldlen));
519				if (i)
520					error = i;
521			}
522		}
523		if ((error == ENOMEM || !error ) && oldp) {
524			i = copyout(oldp, old, oldlen);
525			if (i)
526				error = i;
527			free(oldp, M_TEMP);
528		}
529		if (newp)
530			free(newp, M_TEMP);
531		return (error);
532	}
533
534	if (oldp)
535		free(oldp, M_TEMP);
536	if (newp)
537		free(newp, M_TEMP);
538
539	switch (name[0]) {
540	case CTL_KERN:
541		fn = kern_sysctl;
542		if (name[1] != KERN_VNODE)      /* XXX */
543			dolock = 0;
544		break;
545	case CTL_HW:
546		fn = hw_sysctl;
547		break;
548	case CTL_VM:
549		fn = vm_sysctl;
550		break;
551	case CTL_NET:
552		fn = net_sysctl;
553		break;
554	case CTL_FS:
555		fn = fs_sysctl;
556		break;
557#ifdef DEBUG
558	case CTL_DEBUG:
559		fn = debug_sysctl;
560		break;
561#endif
562	default:
563		return (EOPNOTSUPP);
564	}
565	if (old != NULL) {
566		if (!useracc(old, oldlen, B_WRITE))
567			return (EFAULT);
568		while (memlock.sl_lock) {
569			memlock.sl_want = 1;
570			(void) tsleep((caddr_t)&memlock, PRIBIO+1, "sysctl", 0);
571			memlock.sl_locked++;
572		}
573		memlock.sl_lock = 1;
574		if (dolock)
575			vslock(old, oldlen);
576		savelen = oldlen;
577	}
578
579
580	error = (*fn)(name + 1, namelen - 1, old, &oldlen,
581	    new, newlen, p);
582
583
584	if (old != NULL) {
585		if (dolock)
586			vsunlock(old, savelen, B_WRITE);
587		memlock.sl_lock = 0;
588		if (memlock.sl_want) {
589			memlock.sl_want = 0;
590			wakeup((caddr_t)&memlock);
591		}
592	}
593#if 0
594	if (error) {
595		printf("SYSCTL_ERROR: ");
596		for(i=0;i<namelen;i++)
597			printf("%d ", name[i]);
598		printf("= %d\n", error);
599	}
600#endif
601	if (error)
602		return (error);
603	if (retval)
604		*retval = oldlen;
605	if (oldlenp) {
606		if (inkernel) {
607			*oldlenp = oldlen;
608		} else {
609			error = copyout(&oldlen, oldlenp, sizeof(oldlen));
610		}
611	}
612	return (error);
613}
614
615/*
616 * Attributes stored in the kernel.
617 */
618int securelevel = -1;
619
620/*
621 * kernel related system variables.
622 */
623int
624kern_sysctl(name, namelen, oldp, oldlenp, newp, newlen, p)
625	int *name;
626	u_int namelen;
627	void *oldp;
628	size_t *oldlenp;
629	void *newp;
630	size_t newlen;
631	struct proc *p;
632{
633	int error, level;
634	dev_t ndumpdev;
635
636	/* all sysctl names at this level are terminal */
637	if (namelen != 1 && !(name[0] == KERN_PROC || name[0] == KERN_PROF
638			      || name[0] == KERN_NTP_PLL))
639		return (ENOTDIR);		/* overloaded */
640
641	switch (name[0]) {
642
643	case KERN_SECURELVL:
644		level = securelevel;
645		if ((error = sysctl_int(oldp, oldlenp, newp, newlen, &level)) ||
646		    newp == NULL)
647			return (error);
648		if (level < securelevel && p->p_pid != 1)
649			return (EPERM);
650		securelevel = level;
651		return (0);
652	case KERN_VNODE:
653		return (sysctl_vnode(oldp, oldlenp));
654	case KERN_PROC:
655		return (sysctl_doproc(name + 1, namelen - 1, oldp, oldlenp));
656	case KERN_FILE:
657		return (sysctl_file(oldp, oldlenp));
658#ifdef GPROF
659	case KERN_PROF:
660		return (sysctl_doprof(name + 1, namelen - 1, oldp, oldlenp,
661		    newp, newlen));
662#endif
663	case KERN_NTP_PLL:
664		return (ntp_sysctl(name + 1, namelen - 1, oldp, oldlenp,
665				   newp, newlen, p));
666	case KERN_DUMPDEV:
667		ndumpdev = dumpdev;
668		error = sysctl_struct(oldp, oldlenp, newp, newlen, &ndumpdev,
669				      sizeof ndumpdev);
670		if (!error && ndumpdev != dumpdev) {
671			error = setdumpdev(ndumpdev);
672		}
673		return error;
674	default:
675		return (EOPNOTSUPP);
676	}
677	/* NOTREACHED */
678}
679
680/*
681 * hardware related system variables.
682 */
683int
684hw_sysctl(name, namelen, oldp, oldlenp, newp, newlen, p)
685	int *name;
686	u_int namelen;
687	void *oldp;
688	size_t *oldlenp;
689	void *newp;
690	size_t newlen;
691	struct proc *p;
692{
693	/* almost all sysctl names at this level are terminal */
694	if (namelen != 1 && name[0] != HW_DEVCONF)
695		return (ENOTDIR);		/* overloaded */
696
697	switch (name[0]) {
698	case HW_PHYSMEM:
699		return (sysctl_rdint(oldp, oldlenp, newp, ctob(physmem)));
700	case HW_USERMEM:
701		return (sysctl_rdint(oldp, oldlenp, newp,
702		    ctob(physmem - cnt.v_wire_count)));
703	case HW_DEVCONF:
704		return (dev_sysctl(name + 1, namelen - 1, oldp, oldlenp,
705				   newp, newlen, p));
706	default:
707		return (EOPNOTSUPP);
708	}
709	/* NOTREACHED */
710}
711
712#ifdef DEBUG
713/*
714 * Debugging related system variables.
715 */
716struct ctldebug debug0, debug1, debug2, debug3, debug4;
717struct ctldebug debug5, debug6, debug7, debug8, debug9;
718struct ctldebug debug10, debug11, debug12, debug13, debug14;
719struct ctldebug debug15, debug16, debug17, debug18, debug19;
720static struct ctldebug *debugvars[CTL_DEBUG_MAXID] = {
721	&debug0, &debug1, &debug2, &debug3, &debug4,
722	&debug5, &debug6, &debug7, &debug8, &debug9,
723	&debug10, &debug11, &debug12, &debug13, &debug14,
724	&debug15, &debug16, &debug17, &debug18, &debug19,
725};
726static int
727debug_sysctl(name, namelen, oldp, oldlenp, newp, newlen, p)
728	int *name;
729	u_int namelen;
730	void *oldp;
731	size_t *oldlenp;
732	void *newp;
733	size_t newlen;
734	struct proc *p;
735{
736	struct ctldebug *cdp;
737
738	/* all sysctl names at this level are name and field */
739	if (namelen != 2)
740		return (ENOTDIR);		/* overloaded */
741	cdp = debugvars[name[0]];
742	if (cdp->debugname == 0)
743		return (EOPNOTSUPP);
744	switch (name[1]) {
745	case CTL_DEBUG_NAME:
746		return (sysctl_rdstring(oldp, oldlenp, newp, cdp->debugname));
747	case CTL_DEBUG_VALUE:
748		return (sysctl_int(oldp, oldlenp, newp, newlen, cdp->debugvar));
749	default:
750		return (EOPNOTSUPP);
751	}
752	/* NOTREACHED */
753}
754#endif /* DEBUG */
755
756/*
757 * Validate parameters and get old / set new parameters
758 * for an integer-valued sysctl function.
759 */
760int
761sysctl_int(oldp, oldlenp, newp, newlen, valp)
762	void *oldp;
763	size_t *oldlenp;
764	void *newp;
765	size_t newlen;
766	int *valp;
767{
768	int error = 0;
769
770	if (oldp && *oldlenp < sizeof(int))
771		return (ENOMEM);
772	if (newp && newlen != sizeof(int))
773		return (EINVAL);
774	*oldlenp = sizeof(int);
775	if (oldp)
776		error = copyout(valp, oldp, sizeof(int));
777	if (error == 0 && newp)
778		error = copyin(newp, valp, sizeof(int));
779	return (error);
780}
781
782/*
783 * As above, but read-only.
784 */
785int
786sysctl_rdint(oldp, oldlenp, newp, val)
787	void *oldp;
788	size_t *oldlenp;
789	void *newp;
790	int val;
791{
792	int error = 0;
793
794	if (oldp && *oldlenp < sizeof(int))
795		return (ENOMEM);
796	if (newp)
797		return (EPERM);
798	*oldlenp = sizeof(int);
799	if (oldp)
800		error = copyout((caddr_t)&val, oldp, sizeof(int));
801	return (error);
802}
803
804/*
805 * Validate parameters and get old / set new parameters
806 * for a string-valued sysctl function.
807 */
808int
809sysctl_string(oldp, oldlenp, newp, newlen, str, maxlen)
810	void *oldp;
811	size_t *oldlenp;
812	void *newp;
813	size_t newlen;
814	char *str;
815	int maxlen;
816{
817	int len, error = 0, rval = 0;
818
819	len = strlen(str) + 1;
820	if (oldp && *oldlenp < len) {
821		len = *oldlenp;
822		rval = ENOMEM;
823	}
824	if (newp && newlen >= maxlen)
825		return (EINVAL);
826	if (oldp) {
827		*oldlenp = len;
828		error = copyout(str, oldp, len);
829		if (error)
830			rval = error;
831	}
832	if ((error == 0 || error == ENOMEM) && newp) {
833		error = copyin(newp, str, newlen);
834		if (error)
835			rval = error;
836		str[newlen] = 0;
837	}
838	return (rval);
839}
840
841/*
842 * As above, but read-only.
843 */
844int
845sysctl_rdstring(oldp, oldlenp, newp, str)
846	void *oldp;
847	size_t *oldlenp;
848	void *newp;
849	char *str;
850{
851	int len, error = 0, rval = 0;
852
853	len = strlen(str) + 1;
854	if (oldp && *oldlenp < len) {
855		len = *oldlenp;
856		rval = ENOMEM;
857	}
858	if (newp)
859		return (EPERM);
860	*oldlenp = len;
861	if (oldp)
862		error = copyout(str, oldp, len);
863		if (error)
864			rval = error;
865	return (rval);
866}
867
868/*
869 * Validate parameters and get old / set new parameters
870 * for a structure oriented sysctl function.
871 */
872int
873sysctl_struct(oldp, oldlenp, newp, newlen, sp, len)
874	void *oldp;
875	size_t *oldlenp;
876	void *newp;
877	size_t newlen;
878	void *sp;
879	int len;
880{
881	int error = 0;
882
883	if (oldp && *oldlenp < len)
884		return (ENOMEM);
885	if (newp && newlen > len)
886		return (EINVAL);
887	if (oldp) {
888		*oldlenp = len;
889		error = copyout(sp, oldp, len);
890	}
891	if (error == 0 && newp)
892		error = copyin(newp, sp, len);
893	return (error);
894}
895
896/*
897 * Validate parameters and get old parameters
898 * for a structure oriented sysctl function.
899 */
900int
901sysctl_rdstruct(oldp, oldlenp, newp, sp, len)
902	void *oldp;
903	size_t *oldlenp;
904	void *newp, *sp;
905	int len;
906{
907	int error = 0;
908
909	if (oldp && *oldlenp < len)
910		return (ENOMEM);
911	if (newp)
912		return (EPERM);
913	*oldlenp = len;
914	if (oldp)
915		error = copyout(sp, oldp, len);
916	return (error);
917}
918
919/*
920 * Get file structures.
921 */
922int
923sysctl_file(where, sizep)
924	char *where;
925	size_t *sizep;
926{
927	int buflen, error;
928	struct file *fp;
929	char *start = where;
930
931	buflen = *sizep;
932	if (where == NULL) {
933		/*
934		 * overestimate by 10 files
935		 */
936		*sizep = sizeof(filehead) + (nfiles + 10) * sizeof(struct file);
937		return (0);
938	}
939
940	/*
941	 * first copyout filehead
942	 */
943	if (buflen < sizeof(filehead)) {
944		*sizep = 0;
945		return (0);
946	}
947	error = copyout((caddr_t)&filehead, where, sizeof(filehead));
948	if (error)
949		return (error);
950	buflen -= sizeof(filehead);
951	where += sizeof(filehead);
952
953	/*
954	 * followed by an array of file structures
955	 */
956	for (fp = filehead; fp != NULL; fp = fp->f_filef) {
957		if (buflen < sizeof(struct file)) {
958			*sizep = where - start;
959			return (ENOMEM);
960		}
961		error = copyout((caddr_t)fp, where, sizeof (struct file));
962		if (error)
963			return (error);
964		buflen -= sizeof(struct file);
965		where += sizeof(struct file);
966	}
967	*sizep = where - start;
968	return (0);
969}
970
971/*
972 * try over estimating by 5 procs
973 */
974#define KERN_PROCSLOP	(5 * sizeof (struct kinfo_proc))
975
976int
977sysctl_doproc(name, namelen, where, sizep)
978	int *name;
979	u_int namelen;
980	char *where;
981	size_t *sizep;
982{
983	register struct proc *p;
984	register struct kinfo_proc *dp = (struct kinfo_proc *)where;
985	register int needed = 0;
986	int buflen = where != NULL ? *sizep : 0;
987	int doingzomb;
988	struct eproc eproc;
989	int error = 0;
990
991	if (namelen != 2 && !(namelen == 1 && name[0] == KERN_PROC_ALL))
992		return (EINVAL);
993	p = (struct proc *)allproc;
994	doingzomb = 0;
995again:
996	for (; p != NULL; p = p->p_next) {
997		/*
998		 * Skip embryonic processes.
999		 */
1000		if (p->p_stat == SIDL)
1001			continue;
1002		/*
1003		 * TODO - make more efficient (see notes below).
1004		 * do by session.
1005		 */
1006		switch (name[0]) {
1007
1008		case KERN_PROC_PID:
1009			/* could do this with just a lookup */
1010			if (p->p_pid != (pid_t)name[1])
1011				continue;
1012			break;
1013
1014		case KERN_PROC_PGRP:
1015			/* could do this by traversing pgrp */
1016			if (p->p_pgrp == NULL || p->p_pgrp->pg_id != (pid_t)name[1])
1017				continue;
1018			break;
1019
1020		case KERN_PROC_TTY:
1021			if ((p->p_flag & P_CONTROLT) == 0 ||
1022			    p->p_session == NULL ||
1023			    p->p_session->s_ttyp == NULL ||
1024			    p->p_session->s_ttyp->t_dev != (dev_t)name[1])
1025				continue;
1026			break;
1027
1028		case KERN_PROC_UID:
1029			if (p->p_ucred == NULL || p->p_ucred->cr_uid != (uid_t)name[1])
1030				continue;
1031			break;
1032
1033		case KERN_PROC_RUID:
1034			if (p->p_ucred == NULL || p->p_cred->p_ruid != (uid_t)name[1])
1035				continue;
1036			break;
1037		}
1038		if (buflen >= sizeof(struct kinfo_proc)) {
1039			fill_eproc(p, &eproc);
1040			error = copyout((caddr_t)p, &dp->kp_proc,
1041			    sizeof(struct proc));
1042			if (error)
1043				return (error);
1044			error = copyout((caddr_t)&eproc, &dp->kp_eproc,
1045			    sizeof(eproc));
1046			if (error)
1047				return (error);
1048			dp++;
1049			buflen -= sizeof(struct kinfo_proc);
1050		}
1051		needed += sizeof(struct kinfo_proc);
1052	}
1053	if (doingzomb == 0) {
1054		p = zombproc;
1055		doingzomb++;
1056		goto again;
1057	}
1058	if (where != NULL) {
1059		*sizep = (caddr_t)dp - where;
1060		if (needed > *sizep)
1061			return (ENOMEM);
1062	} else {
1063		needed += KERN_PROCSLOP;
1064		*sizep = needed;
1065	}
1066	return (0);
1067}
1068
1069/*
1070 * Fill in an eproc structure for the specified process.
1071 */
1072void
1073fill_eproc(p, ep)
1074	register struct proc *p;
1075	register struct eproc *ep;
1076{
1077	register struct tty *tp;
1078
1079	bzero(ep, sizeof(*ep));
1080
1081	ep->e_paddr = p;
1082	if (p->p_cred) {
1083		ep->e_pcred = *p->p_cred;
1084		if (p->p_ucred)
1085			ep->e_ucred = *p->p_ucred;
1086	}
1087	if (p->p_stat != SIDL && p->p_stat != SZOMB && p->p_vmspace != NULL) {
1088		register struct vmspace *vm = p->p_vmspace;
1089
1090#ifdef pmap_resident_count
1091		ep->e_vm.vm_rssize = pmap_resident_count(&vm->vm_pmap); /*XXX*/
1092#else
1093		ep->e_vm.vm_rssize = vm->vm_rssize;
1094#endif
1095		ep->e_vm.vm_tsize = vm->vm_tsize;
1096		ep->e_vm.vm_dsize = vm->vm_dsize;
1097		ep->e_vm.vm_ssize = vm->vm_ssize;
1098#ifndef sparc
1099		ep->e_vm.vm_pmap = vm->vm_pmap;
1100#endif
1101	}
1102	if (p->p_pptr)
1103		ep->e_ppid = p->p_pptr->p_pid;
1104	if (p->p_pgrp) {
1105		ep->e_sess = p->p_pgrp->pg_session;
1106		ep->e_pgid = p->p_pgrp->pg_id;
1107		ep->e_jobc = p->p_pgrp->pg_jobc;
1108	}
1109	if ((p->p_flag & P_CONTROLT) &&
1110	    (ep->e_sess != NULL) &&
1111	    ((tp = ep->e_sess->s_ttyp) != NULL)) {
1112		ep->e_tdev = tp->t_dev;
1113		ep->e_tpgid = tp->t_pgrp ? tp->t_pgrp->pg_id : NO_PID;
1114		ep->e_tsess = tp->t_session;
1115	} else
1116		ep->e_tdev = NODEV;
1117	if (ep->e_sess && ep->e_sess->s_ttyvp)
1118		ep->e_flag = EPROC_CTTY;
1119	if (SESS_LEADER(p))
1120		ep->e_flag |= EPROC_SLEADER;
1121	if (p->p_wmesg) {
1122		strncpy(ep->e_wmesg, p->p_wmesg, WMESGLEN);
1123		ep->e_wmesg[WMESGLEN] = 0;
1124	}
1125}
1126
1127#ifdef COMPAT_43
1128#include <sys/socket.h>
1129#define	KINFO_PROC		(0<<8)
1130#define	KINFO_RT		(1<<8)
1131#define	KINFO_VNODE		(2<<8)
1132#define	KINFO_FILE		(3<<8)
1133#define	KINFO_METER		(4<<8)
1134#define	KINFO_LOADAVG		(5<<8)
1135#define	KINFO_CLOCKRATE		(6<<8)
1136
1137/* Non-standard BSDI extension - only present on their 4.3 net-2 releases */
1138#define	KINFO_BSDI_SYSINFO	(101<<8)
1139
1140/*
1141 * XXX this is bloat, but I hope it's better here than on the potentially
1142 * limited kernel stack...  -Peter
1143 */
1144
1145struct {
1146	int	bsdi_machine;		/* "i386" on BSD/386 */
1147/*      ^^^ this is an offset to the string, relative to the struct start */
1148	char	*pad0;
1149	long	pad1;
1150	long	pad2;
1151	long	pad3;
1152	u_long	pad4;
1153	u_long	pad5;
1154	u_long	pad6;
1155
1156	int	bsdi_ostype;		/* "BSD/386" on BSD/386 */
1157	int	bsdi_osrelease;		/* "1.1" on BSD/386 */
1158	long	pad7;
1159	long	pad8;
1160	char	*pad9;
1161
1162	long	pad10;
1163	long	pad11;
1164	int	pad12;
1165	long	pad13;
1166	quad_t	pad14;
1167	long	pad15;
1168
1169	struct	timeval pad16;
1170	/* we dont set this, because BSDI's uname used gethostname() instead */
1171	int	bsdi_hostname;		/* hostname on BSD/386 */
1172
1173	/* the actual string data is appended here */
1174
1175} bsdi_si;
1176/*
1177 * this data is appended to the end of the bsdi_si structure during copyout.
1178 * The "char *" offsets are relative to the base of the bsdi_si struct.
1179 * This contains "FreeBSD\02.0-BUILT-nnnnnn\0i386\0", and these strings
1180 * should not exceed the length of the buffer here... (or else!! :-)
1181 */
1182char bsdi_strings[80];	/* It had better be less than this! */
1183
1184struct getkerninfo_args {
1185	int	op;
1186	char	*where;
1187	int	*size;
1188	int	arg;
1189};
1190
1191int
1192ogetkerninfo(p, uap, retval)
1193	struct proc *p;
1194	register struct getkerninfo_args *uap;
1195	int *retval;
1196{
1197	int error, name[6];
1198	u_int size;
1199
1200	switch (uap->op & 0xff00) {
1201
1202	case KINFO_RT:
1203		name[0] = CTL_NET;
1204		name[1] = PF_ROUTE;
1205		name[2] = 0;
1206		name[3] = (uap->op & 0xff0000) >> 16;
1207		name[4] = uap->op & 0xff;
1208		name[5] = uap->arg;
1209		error = userland_sysctl(p, name, 6, uap->where, uap->size,
1210			0, 0, 0, 0);
1211		break;
1212
1213	case KINFO_VNODE:
1214		name[0] = CTL_KERN;
1215		name[1] = KERN_VNODE;
1216		error = userland_sysctl(p, name, 2, uap->where, uap->size,
1217			0, 0, 0, 0);
1218		break;
1219
1220	case KINFO_PROC:
1221		name[0] = CTL_KERN;
1222		name[1] = KERN_PROC;
1223		name[2] = uap->op & 0xff;
1224		name[3] = uap->arg;
1225		error = userland_sysctl(p, name, 4, uap->where, uap->size,
1226			0, 0, 0, 0);
1227		break;
1228
1229	case KINFO_FILE:
1230		name[0] = CTL_KERN;
1231		name[1] = KERN_FILE;
1232		error = userland_sysctl(p, name, 2, uap->where, uap->size,
1233			0, 0, 0, 0);
1234		break;
1235
1236	case KINFO_METER:
1237		name[0] = CTL_VM;
1238		name[1] = VM_METER;
1239		error = userland_sysctl(p, name, 2, uap->where, uap->size,
1240			0, 0, 0, 0);
1241		break;
1242
1243	case KINFO_LOADAVG:
1244		name[0] = CTL_VM;
1245		name[1] = VM_LOADAVG;
1246		error = userland_sysctl(p, name, 2, uap->where, uap->size,
1247			0, 0, 0, 0);
1248		break;
1249
1250	case KINFO_CLOCKRATE:
1251		name[0] = CTL_KERN;
1252		name[1] = KERN_CLOCKRATE;
1253		error = userland_sysctl(p, name, 2, uap->where, uap->size,
1254			0, 0, 0, 0);
1255		break;
1256
1257	case KINFO_BSDI_SYSINFO: {
1258		/*
1259		 * this is pretty crude, but it's just enough for uname()
1260		 * from BSDI's 1.x libc to work.
1261		 *
1262		 * In particular, it doesn't return the same results when
1263		 * the supplied buffer is too small.  BSDI's version apparently
1264		 * will return the amount copied, and set the *size to how
1265		 * much was needed.  The emulation framework here isn't capable
1266		 * of that, so we just set both to the amount copied.
1267		 * BSDI's 2.x product apparently fails with ENOMEM in this
1268		 * scenario.
1269		 */
1270
1271		u_int needed;
1272		u_int left;
1273		char *s;
1274
1275		bzero((char *)&bsdi_si, sizeof(bsdi_si));
1276		bzero(bsdi_strings, sizeof(bsdi_strings));
1277
1278		s = bsdi_strings;
1279
1280		bsdi_si.bsdi_ostype = (s - bsdi_strings) + sizeof(bsdi_si);
1281		strcpy(s, ostype);
1282		s += strlen(s) + 1;
1283
1284		bsdi_si.bsdi_osrelease = (s - bsdi_strings) + sizeof(bsdi_si);
1285		strcpy(s, osrelease);
1286		s += strlen(s) + 1;
1287
1288		bsdi_si.bsdi_machine = (s - bsdi_strings) + sizeof(bsdi_si);
1289		strcpy(s, machine);
1290		s += strlen(s) + 1;
1291
1292		needed = sizeof(bsdi_si) + (s - bsdi_strings);
1293
1294		if (uap->where == NULL) {
1295			/* process is asking how much buffer to supply.. */
1296			size = needed;
1297			error = 0;
1298			break;
1299		}
1300
1301
1302		/* if too much buffer supplied, trim it down */
1303		if (size > needed)
1304			size = needed;
1305
1306		/* how much of the buffer is remaining */
1307		left = size;
1308
1309		if ((error = copyout((char *)&bsdi_si, uap->where, left)) != 0)
1310			break;
1311
1312		/* is there any point in continuing? */
1313		if (left > sizeof(bsdi_si)) {
1314			left -= sizeof(bsdi_si);
1315			error = copyout(&bsdi_strings,
1316					uap->where + sizeof(bsdi_si), left);
1317		}
1318		break;
1319	}
1320
1321	default:
1322		return (EOPNOTSUPP);
1323	}
1324	if (error)
1325		return (error);
1326	*retval = size;
1327	if (uap->size)
1328		error = copyout((caddr_t)&size, (caddr_t)uap->size,
1329		    sizeof(size));
1330	return (error);
1331}
1332#endif /* COMPAT_43 */
1333