init_sysctl.c revision 1.136
1/*	$NetBSD: init_sysctl.c,v 1.136 2008/04/29 18:13:24 ad Exp $ */
2
3/*-
4 * Copyright (c) 2003, 2007, 2008 The NetBSD Foundation, Inc.
5 * All rights reserved.
6 *
7 * This code is derived from software contributed to The NetBSD Foundation
8 * by Andrew Brown, and by Andrew Doran.
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 *
19 * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS
20 * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
21 * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
22 * PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS
23 * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
24 * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
25 * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
26 * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
27 * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
28 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
29 * POSSIBILITY OF SUCH DAMAGE.
30 */
31
32#include <sys/cdefs.h>
33__KERNEL_RCSID(0, "$NetBSD: init_sysctl.c,v 1.136 2008/04/29 18:13:24 ad Exp $");
34
35#include "opt_sysv.h"
36#include "opt_posix.h"
37#include "opt_compat_netbsd32.h"
38#include "pty.h"
39#include "rnd.h"
40
41#include <sys/types.h>
42#include <sys/param.h>
43#include <sys/sysctl.h>
44#include <sys/cpu.h>
45#include <sys/errno.h>
46#include <sys/systm.h>
47#include <sys/kernel.h>
48#include <sys/unistd.h>
49#include <sys/disklabel.h>
50#include <sys/rnd.h>
51#include <sys/vnode.h>
52#include <sys/mount.h>
53#include <sys/namei.h>
54#include <sys/msgbuf.h>
55#include <dev/cons.h>
56#include <sys/socketvar.h>
57#include <sys/file.h>
58#include <sys/filedesc.h>
59#include <sys/tty.h>
60#include <sys/malloc.h>
61#include <sys/resource.h>
62#include <sys/resourcevar.h>
63#include <sys/exec.h>
64#include <sys/conf.h>
65#include <sys/device.h>
66#include <sys/stat.h>
67#include <sys/kauth.h>
68#include <sys/ktrace.h>
69
70#ifdef COMPAT_NETBSD32
71#include <compat/netbsd32/netbsd32.h>
72#endif
73
74#include <sys/cpu.h>
75
76/* XXX this should not be here */
77int security_setidcore_dump;
78char security_setidcore_path[MAXPATHLEN] = "/var/crash/%n.core";
79uid_t security_setidcore_owner = 0;
80gid_t security_setidcore_group = 0;
81mode_t security_setidcore_mode = (S_IRUSR|S_IWUSR);
82
83static const u_int sysctl_flagmap[] = {
84	PK_ADVLOCK, P_ADVLOCK,
85	PK_EXEC, P_EXEC,
86	PK_NOCLDWAIT, P_NOCLDWAIT,
87	PK_32, P_32,
88	PK_CLDSIGIGN, P_CLDSIGIGN,
89	PK_SUGID, P_SUGID,
90	0
91};
92
93static const u_int sysctl_sflagmap[] = {
94	PS_NOCLDSTOP, P_NOCLDSTOP,
95	PS_PPWAIT, P_PPWAIT,
96	PS_WEXIT, P_WEXIT,
97	PS_STOPFORK, P_STOPFORK,
98	PS_STOPEXEC, P_STOPEXEC,
99	PS_STOPEXIT, P_STOPEXIT,
100	0
101};
102
103static const u_int sysctl_slflagmap[] = {
104	PSL_TRACED, P_TRACED,
105	PSL_FSTRACE, P_FSTRACE,
106	PSL_CHTRACED, P_CHTRACED,
107	PSL_SYSCALL, P_SYSCALL,
108	0
109};
110
111static const u_int sysctl_lflagmap[] = {
112	PL_CONTROLT, P_CONTROLT,
113	0
114};
115
116static const u_int sysctl_stflagmap[] = {
117	PST_PROFIL, P_PROFIL,
118	0
119
120};
121
122static const u_int sysctl_lwpflagmap[] = {
123	LW_INMEM, P_INMEM,
124	LW_SINTR, P_SINTR,
125	LW_SYSTEM, P_SYSTEM,
126	0
127};
128
129static const u_int sysctl_lwpprflagmap[] = {
130	LPR_DETACHED, L_DETACHED,
131	0
132};
133
134/*
135 * try over estimating by 5 procs/lwps
136 */
137#define KERN_PROCSLOP	(5 * sizeof(struct kinfo_proc))
138#define KERN_LWPSLOP	(5 * sizeof(struct kinfo_lwp))
139
140static int dcopyout(struct lwp *, const void *, void *, size_t);
141
142static int
143dcopyout(struct lwp *l, const void *kaddr, void *uaddr, size_t len)
144{
145	int error;
146
147	error = copyout(kaddr, uaddr, len);
148	ktrmibio(-1, UIO_READ, uaddr, len, error);
149
150	return error;
151}
152
153#ifdef DIAGNOSTIC
154static int sysctl_kern_trigger_panic(SYSCTLFN_PROTO);
155#endif
156static int sysctl_kern_maxvnodes(SYSCTLFN_PROTO);
157static int sysctl_kern_rtc_offset(SYSCTLFN_PROTO);
158static int sysctl_kern_maxproc(SYSCTLFN_PROTO);
159static int sysctl_kern_hostid(SYSCTLFN_PROTO);
160static int sysctl_setlen(SYSCTLFN_PROTO);
161static int sysctl_kern_clockrate(SYSCTLFN_PROTO);
162static int sysctl_kern_file(SYSCTLFN_PROTO);
163static int sysctl_msgbuf(SYSCTLFN_PROTO);
164static int sysctl_kern_defcorename(SYSCTLFN_PROTO);
165static int sysctl_kern_cptime(SYSCTLFN_PROTO);
166#if NPTY > 0
167static int sysctl_kern_maxptys(SYSCTLFN_PROTO);
168#endif /* NPTY > 0 */
169static int sysctl_kern_sbmax(SYSCTLFN_PROTO);
170static int sysctl_kern_urnd(SYSCTLFN_PROTO);
171static int sysctl_kern_arnd(SYSCTLFN_PROTO);
172static int sysctl_kern_lwp(SYSCTLFN_PROTO);
173static int sysctl_kern_forkfsleep(SYSCTLFN_PROTO);
174static int sysctl_kern_root_partition(SYSCTLFN_PROTO);
175static int sysctl_kern_drivers(SYSCTLFN_PROTO);
176static int sysctl_kern_file2(SYSCTLFN_PROTO);
177static int sysctl_security_setidcore(SYSCTLFN_PROTO);
178static int sysctl_security_setidcorename(SYSCTLFN_PROTO);
179static int sysctl_kern_cpid(SYSCTLFN_PROTO);
180static int sysctl_doeproc(SYSCTLFN_PROTO);
181static int sysctl_kern_proc_args(SYSCTLFN_PROTO);
182static int sysctl_hw_usermem(SYSCTLFN_PROTO);
183static int sysctl_hw_cnmagic(SYSCTLFN_PROTO);
184
185static u_int sysctl_map_flags(const u_int *, u_int);
186static void fill_kproc2(struct proc *, struct kinfo_proc2 *, bool);
187static void fill_lwp(struct lwp *l, struct kinfo_lwp *kl);
188static void fill_file(struct kinfo_file *, const file_t *, const fdfile_t *,
189		      int, pid_t);
190
191/*
192 * ********************************************************************
193 * section 1: setup routines
194 * ********************************************************************
195 * These functions are stuffed into a link set for sysctl setup
196 * functions. They're never called or referenced from anywhere else.
197 * ********************************************************************
198 */
199
200/*
201 * sets up the base nodes...
202 */
203SYSCTL_SETUP(sysctl_root_setup, "sysctl base setup")
204{
205
206	sysctl_createv(clog, 0, NULL, NULL,
207		       CTLFLAG_PERMANENT,
208		       CTLTYPE_NODE, "kern",
209		       SYSCTL_DESCR("High kernel"),
210		       NULL, 0, NULL, 0,
211		       CTL_KERN, CTL_EOL);
212	sysctl_createv(clog, 0, NULL, NULL,
213		       CTLFLAG_PERMANENT,
214		       CTLTYPE_NODE, "vm",
215		       SYSCTL_DESCR("Virtual memory"),
216		       NULL, 0, NULL, 0,
217		       CTL_VM, CTL_EOL);
218	sysctl_createv(clog, 0, NULL, NULL,
219		       CTLFLAG_PERMANENT,
220		       CTLTYPE_NODE, "vfs",
221		       SYSCTL_DESCR("Filesystem"),
222		       NULL, 0, NULL, 0,
223		       CTL_VFS, CTL_EOL);
224	sysctl_createv(clog, 0, NULL, NULL,
225		       CTLFLAG_PERMANENT,
226		       CTLTYPE_NODE, "net",
227		       SYSCTL_DESCR("Networking"),
228		       NULL, 0, NULL, 0,
229		       CTL_NET, CTL_EOL);
230	sysctl_createv(clog, 0, NULL, NULL,
231		       CTLFLAG_PERMANENT,
232		       CTLTYPE_NODE, "debug",
233		       SYSCTL_DESCR("Debugging"),
234		       NULL, 0, NULL, 0,
235		       CTL_DEBUG, CTL_EOL);
236	sysctl_createv(clog, 0, NULL, NULL,
237		       CTLFLAG_PERMANENT,
238		       CTLTYPE_NODE, "hw",
239		       SYSCTL_DESCR("Generic CPU, I/O"),
240		       NULL, 0, NULL, 0,
241		       CTL_HW, CTL_EOL);
242	sysctl_createv(clog, 0, NULL, NULL,
243		       CTLFLAG_PERMANENT,
244		       CTLTYPE_NODE, "machdep",
245		       SYSCTL_DESCR("Machine dependent"),
246		       NULL, 0, NULL, 0,
247		       CTL_MACHDEP, CTL_EOL);
248	/*
249	 * this node is inserted so that the sysctl nodes in libc can
250	 * operate.
251	 */
252	sysctl_createv(clog, 0, NULL, NULL,
253		       CTLFLAG_PERMANENT,
254		       CTLTYPE_NODE, "user",
255		       SYSCTL_DESCR("User-level"),
256		       NULL, 0, NULL, 0,
257		       CTL_USER, CTL_EOL);
258	sysctl_createv(clog, 0, NULL, NULL,
259		       CTLFLAG_PERMANENT,
260		       CTLTYPE_NODE, "ddb",
261		       SYSCTL_DESCR("In-kernel debugger"),
262		       NULL, 0, NULL, 0,
263		       CTL_DDB, CTL_EOL);
264	sysctl_createv(clog, 0, NULL, NULL,
265		       CTLFLAG_PERMANENT,
266		       CTLTYPE_NODE, "proc",
267		       SYSCTL_DESCR("Per-process"),
268		       NULL, 0, NULL, 0,
269		       CTL_PROC, CTL_EOL);
270	sysctl_createv(clog, 0, NULL, NULL,
271		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
272		       CTLTYPE_NODE, "vendor",
273		       SYSCTL_DESCR("Vendor specific"),
274		       NULL, 0, NULL, 0,
275		       CTL_VENDOR, CTL_EOL);
276	sysctl_createv(clog, 0, NULL, NULL,
277		       CTLFLAG_PERMANENT,
278		       CTLTYPE_NODE, "emul",
279		       SYSCTL_DESCR("Emulation settings"),
280		       NULL, 0, NULL, 0,
281		       CTL_EMUL, CTL_EOL);
282	sysctl_createv(clog, 0, NULL, NULL,
283		       CTLFLAG_PERMANENT,
284		       CTLTYPE_NODE, "security",
285		       SYSCTL_DESCR("Security"),
286		       NULL, 0, NULL, 0,
287		       CTL_SECURITY, CTL_EOL);
288}
289
290/*
291 * this setup routine is a replacement for kern_sysctl()
292 */
293SYSCTL_SETUP(sysctl_kern_setup, "sysctl kern subtree setup")
294{
295	extern int kern_logsigexit;	/* defined in kern/kern_sig.c */
296	extern fixpt_t ccpu;		/* defined in kern/kern_synch.c */
297	extern int dumponpanic;		/* defined in kern/subr_prf.c */
298	const struct sysctlnode *rnode;
299
300	sysctl_createv(clog, 0, NULL, NULL,
301		       CTLFLAG_PERMANENT,
302		       CTLTYPE_NODE, "kern", NULL,
303		       NULL, 0, NULL, 0,
304		       CTL_KERN, CTL_EOL);
305
306	sysctl_createv(clog, 0, NULL, NULL,
307		       CTLFLAG_PERMANENT,
308		       CTLTYPE_STRING, "ostype",
309		       SYSCTL_DESCR("Operating system type"),
310		       NULL, 0, &ostype, 0,
311		       CTL_KERN, KERN_OSTYPE, CTL_EOL);
312	sysctl_createv(clog, 0, NULL, NULL,
313		       CTLFLAG_PERMANENT,
314		       CTLTYPE_STRING, "osrelease",
315		       SYSCTL_DESCR("Operating system release"),
316		       NULL, 0, &osrelease, 0,
317		       CTL_KERN, KERN_OSRELEASE, CTL_EOL);
318	sysctl_createv(clog, 0, NULL, NULL,
319		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
320		       CTLTYPE_INT, "osrevision",
321		       SYSCTL_DESCR("Operating system revision"),
322		       NULL, __NetBSD_Version__, NULL, 0,
323		       CTL_KERN, KERN_OSREV, CTL_EOL);
324	sysctl_createv(clog, 0, NULL, NULL,
325		       CTLFLAG_PERMANENT,
326		       CTLTYPE_STRING, "version",
327		       SYSCTL_DESCR("Kernel version"),
328		       NULL, 0, &version, 0,
329		       CTL_KERN, KERN_VERSION, CTL_EOL);
330	sysctl_createv(clog, 0, NULL, NULL,
331		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
332		       CTLTYPE_INT, "maxvnodes",
333		       SYSCTL_DESCR("Maximum number of vnodes"),
334		       sysctl_kern_maxvnodes, 0, NULL, 0,
335		       CTL_KERN, KERN_MAXVNODES, CTL_EOL);
336	sysctl_createv(clog, 0, NULL, NULL,
337		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
338		       CTLTYPE_INT, "maxproc",
339		       SYSCTL_DESCR("Maximum number of simultaneous processes"),
340		       sysctl_kern_maxproc, 0, NULL, 0,
341		       CTL_KERN, KERN_MAXPROC, CTL_EOL);
342	sysctl_createv(clog, 0, NULL, NULL,
343		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
344		       CTLTYPE_INT, "maxfiles",
345		       SYSCTL_DESCR("Maximum number of open files"),
346		       NULL, 0, &maxfiles, 0,
347		       CTL_KERN, KERN_MAXFILES, CTL_EOL);
348	sysctl_createv(clog, 0, NULL, NULL,
349		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
350		       CTLTYPE_INT, "argmax",
351		       SYSCTL_DESCR("Maximum number of bytes of arguments to "
352				    "execve(2)"),
353		       NULL, ARG_MAX, NULL, 0,
354		       CTL_KERN, KERN_ARGMAX, CTL_EOL);
355	sysctl_createv(clog, 0, NULL, NULL,
356		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
357		       CTLTYPE_STRING, "hostname",
358		       SYSCTL_DESCR("System hostname"),
359		       sysctl_setlen, 0, &hostname, MAXHOSTNAMELEN,
360		       CTL_KERN, KERN_HOSTNAME, CTL_EOL);
361	sysctl_createv(clog, 0, NULL, NULL,
362		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE|CTLFLAG_HEX,
363		       CTLTYPE_INT, "hostid",
364		       SYSCTL_DESCR("System host ID number"),
365		       sysctl_kern_hostid, 0, NULL, 0,
366		       CTL_KERN, KERN_HOSTID, CTL_EOL);
367	sysctl_createv(clog, 0, NULL, NULL,
368		       CTLFLAG_PERMANENT,
369		       CTLTYPE_STRUCT, "clockrate",
370		       SYSCTL_DESCR("Kernel clock rates"),
371		       sysctl_kern_clockrate, 0, NULL,
372		       sizeof(struct clockinfo),
373		       CTL_KERN, KERN_CLOCKRATE, CTL_EOL);
374	sysctl_createv(clog, 0, NULL, NULL,
375		       CTLFLAG_PERMANENT,
376		       CTLTYPE_INT, "hardclock_ticks",
377		       SYSCTL_DESCR("Number of hardclock ticks"),
378		       NULL, 0, &hardclock_ticks, sizeof(hardclock_ticks),
379		       CTL_KERN, KERN_HARDCLOCK_TICKS, CTL_EOL);
380	sysctl_createv(clog, 0, NULL, NULL,
381		       CTLFLAG_PERMANENT,
382		       CTLTYPE_STRUCT, "vnode",
383		       SYSCTL_DESCR("System vnode table"),
384		       sysctl_kern_vnode, 0, NULL, 0,
385		       CTL_KERN, KERN_VNODE, CTL_EOL);
386	sysctl_createv(clog, 0, NULL, NULL,
387		       CTLFLAG_PERMANENT,
388		       CTLTYPE_STRUCT, "file",
389		       SYSCTL_DESCR("System open file table"),
390		       sysctl_kern_file, 0, NULL, 0,
391		       CTL_KERN, KERN_FILE, CTL_EOL);
392#ifndef GPROF
393	sysctl_createv(clog, 0, NULL, NULL,
394		       CTLFLAG_PERMANENT,
395		       CTLTYPE_NODE, "profiling",
396		       SYSCTL_DESCR("Profiling information (not available)"),
397		       sysctl_notavail, 0, NULL, 0,
398		       CTL_KERN, KERN_PROF, CTL_EOL);
399#endif
400	sysctl_createv(clog, 0, NULL, NULL,
401		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
402		       CTLTYPE_INT, "posix1version",
403		       SYSCTL_DESCR("Version of ISO/IEC 9945 (POSIX 1003.1) "
404				    "with which the operating system attempts "
405				    "to comply"),
406		       NULL, _POSIX_VERSION, NULL, 0,
407		       CTL_KERN, KERN_POSIX1, CTL_EOL);
408	sysctl_createv(clog, 0, NULL, NULL,
409		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
410		       CTLTYPE_INT, "ngroups",
411		       SYSCTL_DESCR("Maximum number of supplemental groups"),
412		       NULL, NGROUPS_MAX, NULL, 0,
413		       CTL_KERN, KERN_NGROUPS, CTL_EOL);
414	sysctl_createv(clog, 0, NULL, NULL,
415		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
416		       CTLTYPE_INT, "job_control",
417		       SYSCTL_DESCR("Whether job control is available"),
418		       NULL, 1, NULL, 0,
419		       CTL_KERN, KERN_JOB_CONTROL, CTL_EOL);
420	sysctl_createv(clog, 0, NULL, NULL,
421		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
422		       CTLTYPE_INT, "saved_ids",
423		       SYSCTL_DESCR("Whether POSIX saved set-group/user ID is "
424				    "available"), NULL,
425#ifdef _POSIX_SAVED_IDS
426		       1,
427#else /* _POSIX_SAVED_IDS */
428		       0,
429#endif /* _POSIX_SAVED_IDS */
430		       NULL, 0, CTL_KERN, KERN_SAVED_IDS, CTL_EOL);
431	sysctl_createv(clog, 0, NULL, NULL,
432		       CTLFLAG_PERMANENT,
433		       CTLTYPE_STRUCT, "boottime",
434		       SYSCTL_DESCR("System boot time"),
435		       NULL, 0, &boottime, sizeof(boottime),
436		       CTL_KERN, KERN_BOOTTIME, CTL_EOL);
437	sysctl_createv(clog, 0, NULL, NULL,
438		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
439		       CTLTYPE_STRING, "domainname",
440		       SYSCTL_DESCR("YP domain name"),
441		       sysctl_setlen, 0, &domainname, MAXHOSTNAMELEN,
442		       CTL_KERN, KERN_DOMAINNAME, CTL_EOL);
443	sysctl_createv(clog, 0, NULL, NULL,
444		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
445		       CTLTYPE_INT, "maxpartitions",
446		       SYSCTL_DESCR("Maximum number of partitions allowed per "
447				    "disk"),
448		       NULL, MAXPARTITIONS, NULL, 0,
449		       CTL_KERN, KERN_MAXPARTITIONS, CTL_EOL);
450	sysctl_createv(clog, 0, NULL, NULL,
451		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
452		       CTLTYPE_INT, "rawpartition",
453		       SYSCTL_DESCR("Raw partition of a disk"),
454		       NULL, RAW_PART, NULL, 0,
455		       CTL_KERN, KERN_RAWPARTITION, CTL_EOL);
456	sysctl_createv(clog, 0, NULL, NULL,
457		       CTLFLAG_PERMANENT,
458		       CTLTYPE_STRUCT, "timex", NULL,
459		       sysctl_notavail, 0, NULL, 0,
460		       CTL_KERN, KERN_TIMEX, CTL_EOL);
461	sysctl_createv(clog, 0, NULL, NULL,
462		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
463		       CTLTYPE_INT, "rtc_offset",
464		       SYSCTL_DESCR("Offset of real time clock from UTC in "
465				    "minutes"),
466		       sysctl_kern_rtc_offset, 0, &rtc_offset, 0,
467		       CTL_KERN, KERN_RTC_OFFSET, CTL_EOL);
468	sysctl_createv(clog, 0, NULL, NULL,
469		       CTLFLAG_PERMANENT,
470		       CTLTYPE_STRING, "root_device",
471		       SYSCTL_DESCR("Name of the root device"),
472		       sysctl_root_device, 0, NULL, 0,
473		       CTL_KERN, KERN_ROOT_DEVICE, CTL_EOL);
474	sysctl_createv(clog, 0, NULL, NULL,
475		       CTLFLAG_PERMANENT,
476		       CTLTYPE_INT, "msgbufsize",
477		       SYSCTL_DESCR("Size of the kernel message buffer"),
478		       sysctl_msgbuf, 0, NULL, 0,
479		       CTL_KERN, KERN_MSGBUFSIZE, CTL_EOL);
480	sysctl_createv(clog, 0, NULL, NULL,
481		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
482		       CTLTYPE_INT, "fsync",
483		       SYSCTL_DESCR("Whether the POSIX 1003.1b File "
484				    "Synchronization Option is available on "
485				    "this system"),
486		       NULL, 1, NULL, 0,
487		       CTL_KERN, KERN_FSYNC, CTL_EOL);
488	sysctl_createv(clog, 0, NULL, NULL,
489		       CTLFLAG_PERMANENT,
490		       CTLTYPE_NODE, "ipc",
491		       SYSCTL_DESCR("SysV IPC options"),
492		       NULL, 0, NULL, 0,
493		       CTL_KERN, KERN_SYSVIPC, CTL_EOL);
494	sysctl_createv(clog, 0, NULL, NULL,
495		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
496		       CTLTYPE_INT, "sysvmsg",
497		       SYSCTL_DESCR("System V style message support available"),
498		       NULL,
499#ifdef SYSVMSG
500		       1,
501#else /* SYSVMSG */
502		       0,
503#endif /* SYSVMSG */
504		       NULL, 0, CTL_KERN, KERN_SYSVIPC, KERN_SYSVIPC_MSG, CTL_EOL);
505	sysctl_createv(clog, 0, NULL, NULL,
506		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
507		       CTLTYPE_INT, "sysvsem",
508		       SYSCTL_DESCR("System V style semaphore support "
509				    "available"), NULL,
510#ifdef SYSVSEM
511		       1,
512#else /* SYSVSEM */
513		       0,
514#endif /* SYSVSEM */
515		       NULL, 0, CTL_KERN, KERN_SYSVIPC, KERN_SYSVIPC_SEM, CTL_EOL);
516	sysctl_createv(clog, 0, NULL, NULL,
517		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
518		       CTLTYPE_INT, "sysvshm",
519		       SYSCTL_DESCR("System V style shared memory support "
520				    "available"), NULL,
521#ifdef SYSVSHM
522		       1,
523#else /* SYSVSHM */
524		       0,
525#endif /* SYSVSHM */
526		       NULL, 0, CTL_KERN, KERN_SYSVIPC, KERN_SYSVIPC_SHM, CTL_EOL);
527	sysctl_createv(clog, 0, NULL, NULL,
528		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
529		       CTLTYPE_INT, "synchronized_io",
530		       SYSCTL_DESCR("Whether the POSIX 1003.1b Synchronized "
531				    "I/O Option is available on this system"),
532		       NULL, 1, NULL, 0,
533		       CTL_KERN, KERN_SYNCHRONIZED_IO, CTL_EOL);
534	sysctl_createv(clog, 0, NULL, NULL,
535		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
536		       CTLTYPE_INT, "iov_max",
537		       SYSCTL_DESCR("Maximum number of iovec structures per "
538				    "process"),
539		       NULL, IOV_MAX, NULL, 0,
540		       CTL_KERN, KERN_IOV_MAX, CTL_EOL);
541	sysctl_createv(clog, 0, NULL, NULL,
542		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
543		       CTLTYPE_INT, "mapped_files",
544		       SYSCTL_DESCR("Whether the POSIX 1003.1b Memory Mapped "
545				    "Files Option is available on this system"),
546		       NULL, 1, NULL, 0,
547		       CTL_KERN, KERN_MAPPED_FILES, CTL_EOL);
548	sysctl_createv(clog, 0, NULL, NULL,
549		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
550		       CTLTYPE_INT, "memlock",
551		       SYSCTL_DESCR("Whether the POSIX 1003.1b Process Memory "
552				    "Locking Option is available on this "
553				    "system"),
554		       NULL, 1, NULL, 0,
555		       CTL_KERN, KERN_MEMLOCK, CTL_EOL);
556	sysctl_createv(clog, 0, NULL, NULL,
557		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
558		       CTLTYPE_INT, "memlock_range",
559		       SYSCTL_DESCR("Whether the POSIX 1003.1b Range Memory "
560				    "Locking Option is available on this "
561				    "system"),
562		       NULL, 1, NULL, 0,
563		       CTL_KERN, KERN_MEMLOCK_RANGE, CTL_EOL);
564	sysctl_createv(clog, 0, NULL, NULL,
565		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
566		       CTLTYPE_INT, "memory_protection",
567		       SYSCTL_DESCR("Whether the POSIX 1003.1b Memory "
568				    "Protection Option is available on this "
569				    "system"),
570		       NULL, 1, NULL, 0,
571		       CTL_KERN, KERN_MEMORY_PROTECTION, CTL_EOL);
572	sysctl_createv(clog, 0, NULL, NULL,
573		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
574		       CTLTYPE_INT, "login_name_max",
575		       SYSCTL_DESCR("Maximum login name length"),
576		       NULL, LOGIN_NAME_MAX, NULL, 0,
577		       CTL_KERN, KERN_LOGIN_NAME_MAX, CTL_EOL);
578	sysctl_createv(clog, 0, NULL, NULL,
579		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
580		       CTLTYPE_STRING, "defcorename",
581		       SYSCTL_DESCR("Default core file name"),
582		       sysctl_kern_defcorename, 0, defcorename, MAXPATHLEN,
583		       CTL_KERN, KERN_DEFCORENAME, CTL_EOL);
584	sysctl_createv(clog, 0, NULL, NULL,
585		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
586		       CTLTYPE_INT, "logsigexit",
587		       SYSCTL_DESCR("Log process exit when caused by signals"),
588		       NULL, 0, &kern_logsigexit, 0,
589		       CTL_KERN, KERN_LOGSIGEXIT, CTL_EOL);
590	sysctl_createv(clog, 0, NULL, NULL,
591		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
592		       CTLTYPE_INT, "fscale",
593		       SYSCTL_DESCR("Kernel fixed-point scale factor"),
594		       NULL, FSCALE, NULL, 0,
595		       CTL_KERN, KERN_FSCALE, CTL_EOL);
596	sysctl_createv(clog, 0, NULL, NULL,
597		       CTLFLAG_PERMANENT,
598		       CTLTYPE_INT, "ccpu",
599		       SYSCTL_DESCR("Scheduler exponential decay value"),
600		       NULL, 0, &ccpu, 0,
601		       CTL_KERN, KERN_CCPU, CTL_EOL);
602	sysctl_createv(clog, 0, NULL, NULL,
603		       CTLFLAG_PERMANENT,
604		       CTLTYPE_STRUCT, "cp_time",
605		       SYSCTL_DESCR("Clock ticks spent in different CPU states"),
606		       sysctl_kern_cptime, 0, NULL, 0,
607		       CTL_KERN, KERN_CP_TIME, CTL_EOL);
608	sysctl_createv(clog, 0, NULL, NULL,
609		       CTLFLAG_PERMANENT,
610		       CTLTYPE_INT, "msgbuf",
611		       SYSCTL_DESCR("Kernel message buffer"),
612		       sysctl_msgbuf, 0, NULL, 0,
613		       CTL_KERN, KERN_MSGBUF, CTL_EOL);
614	sysctl_createv(clog, 0, NULL, NULL,
615		       CTLFLAG_PERMANENT,
616		       CTLTYPE_STRUCT, "consdev",
617		       SYSCTL_DESCR("Console device"),
618		       sysctl_consdev, 0, NULL, sizeof(dev_t),
619		       CTL_KERN, KERN_CONSDEV, CTL_EOL);
620#if NPTY > 0
621	sysctl_createv(clog, 0, NULL, NULL,
622		       CTLFLAG_PERMANENT,
623		       CTLTYPE_INT, "maxptys",
624		       SYSCTL_DESCR("Maximum number of pseudo-ttys"),
625		       sysctl_kern_maxptys, 0, NULL, 0,
626		       CTL_KERN, KERN_MAXPTYS, CTL_EOL);
627#endif /* NPTY > 0 */
628	sysctl_createv(clog, 0, NULL, NULL,
629		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
630		       CTLTYPE_INT, "maxphys",
631		       SYSCTL_DESCR("Maximum raw I/O transfer size"),
632		       NULL, MAXPHYS, NULL, 0,
633		       CTL_KERN, KERN_MAXPHYS, CTL_EOL);
634	sysctl_createv(clog, 0, NULL, NULL,
635		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
636		       CTLTYPE_INT, "sbmax",
637		       SYSCTL_DESCR("Maximum socket buffer size"),
638		       sysctl_kern_sbmax, 0, NULL, 0,
639		       CTL_KERN, KERN_SBMAX, CTL_EOL);
640	sysctl_createv(clog, 0, NULL, NULL,
641		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
642		       CTLTYPE_INT, "monotonic_clock",
643		       SYSCTL_DESCR("Implementation version of the POSIX "
644				    "1003.1b Monotonic Clock Option"),
645		       /* XXX _POSIX_VERSION */
646		       NULL, _POSIX_MONOTONIC_CLOCK, NULL, 0,
647		       CTL_KERN, KERN_MONOTONIC_CLOCK, CTL_EOL);
648	sysctl_createv(clog, 0, NULL, NULL,
649		       CTLFLAG_PERMANENT,
650		       CTLTYPE_INT, "urandom",
651		       SYSCTL_DESCR("Random integer value"),
652		       sysctl_kern_urnd, 0, NULL, 0,
653		       CTL_KERN, KERN_URND, CTL_EOL);
654	sysctl_createv(clog, 0, NULL, NULL,
655		       CTLFLAG_PERMANENT,
656		       CTLTYPE_INT, "arandom",
657		       SYSCTL_DESCR("n bytes of random data"),
658		       sysctl_kern_arnd, 0, NULL, 0,
659		       CTL_KERN, KERN_ARND, CTL_EOL);
660	sysctl_createv(clog, 0, NULL, NULL,
661		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
662		       CTLTYPE_INT, "labelsector",
663		       SYSCTL_DESCR("Sector number containing the disklabel"),
664		       NULL, LABELSECTOR, NULL, 0,
665		       CTL_KERN, KERN_LABELSECTOR, CTL_EOL);
666	sysctl_createv(clog, 0, NULL, NULL,
667		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
668		       CTLTYPE_INT, "labeloffset",
669		       SYSCTL_DESCR("Offset of the disklabel within the "
670				    "sector"),
671		       NULL, LABELOFFSET, NULL, 0,
672		       CTL_KERN, KERN_LABELOFFSET, CTL_EOL);
673	sysctl_createv(clog, 0, NULL, NULL,
674		       CTLFLAG_PERMANENT,
675		       CTLTYPE_NODE, "lwp",
676		       SYSCTL_DESCR("System-wide LWP information"),
677		       sysctl_kern_lwp, 0, NULL, 0,
678		       CTL_KERN, KERN_LWP, CTL_EOL);
679	sysctl_createv(clog, 0, NULL, NULL,
680		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
681		       CTLTYPE_INT, "forkfsleep",
682		       SYSCTL_DESCR("Milliseconds to sleep on fork failure due "
683				    "to process limits"),
684		       sysctl_kern_forkfsleep, 0, NULL, 0,
685		       CTL_KERN, KERN_FORKFSLEEP, CTL_EOL);
686	sysctl_createv(clog, 0, NULL, NULL,
687		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
688		       CTLTYPE_INT, "posix_threads",
689		       SYSCTL_DESCR("Version of IEEE Std 1003.1 and its "
690				    "Threads option to which the system "
691				    "attempts to conform"),
692		       /* XXX _POSIX_VERSION */
693		       NULL, _POSIX_THREADS, NULL, 0,
694		       CTL_KERN, KERN_POSIX_THREADS, CTL_EOL);
695	sysctl_createv(clog, 0, NULL, NULL,
696		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
697		       CTLTYPE_INT, "posix_semaphores",
698		       SYSCTL_DESCR("Version of IEEE Std 1003.1 and its "
699				    "Semaphores option to which the system "
700				    "attempts to conform"), NULL,
701#ifdef P1003_1B_SEMAPHORE
702		       200112,
703#else /* P1003_1B_SEMAPHORE */
704		       0,
705#endif /* P1003_1B_SEMAPHORE */
706		       NULL, 0, CTL_KERN, KERN_POSIX_SEMAPHORES, CTL_EOL);
707	sysctl_createv(clog, 0, NULL, NULL,
708		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
709		       CTLTYPE_INT, "posix_barriers",
710		       SYSCTL_DESCR("Version of IEEE Std 1003.1 and its "
711				    "Barriers option to which the system "
712				    "attempts to conform"),
713		       /* XXX _POSIX_VERSION */
714		       NULL, _POSIX_BARRIERS, NULL, 0,
715		       CTL_KERN, KERN_POSIX_BARRIERS, CTL_EOL);
716	sysctl_createv(clog, 0, NULL, NULL,
717		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
718		       CTLTYPE_INT, "posix_timers",
719		       SYSCTL_DESCR("Version of IEEE Std 1003.1 and its "
720				    "Timers option to which the system "
721				    "attempts to conform"),
722		       /* XXX _POSIX_VERSION */
723		       NULL, _POSIX_TIMERS, NULL, 0,
724		       CTL_KERN, KERN_POSIX_TIMERS, CTL_EOL);
725	sysctl_createv(clog, 0, NULL, NULL,
726		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
727		       CTLTYPE_INT, "posix_spin_locks",
728		       SYSCTL_DESCR("Version of IEEE Std 1003.1 and its Spin "
729				    "Locks option to which the system attempts "
730				    "to conform"),
731		       /* XXX _POSIX_VERSION */
732		       NULL, _POSIX_SPIN_LOCKS, NULL, 0,
733		       CTL_KERN, KERN_POSIX_SPIN_LOCKS, CTL_EOL);
734	sysctl_createv(clog, 0, NULL, NULL,
735		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
736		       CTLTYPE_INT, "posix_reader_writer_locks",
737		       SYSCTL_DESCR("Version of IEEE Std 1003.1 and its "
738				    "Read-Write Locks option to which the "
739				    "system attempts to conform"),
740		       /* XXX _POSIX_VERSION */
741		       NULL, _POSIX_READER_WRITER_LOCKS, NULL, 0,
742		       CTL_KERN, KERN_POSIX_READER_WRITER_LOCKS, CTL_EOL);
743	sysctl_createv(clog, 0, NULL, NULL,
744		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
745		       CTLTYPE_INT, "dump_on_panic",
746		       SYSCTL_DESCR("Perform a crash dump on system panic"),
747		       NULL, 0, &dumponpanic, 0,
748		       CTL_KERN, KERN_DUMP_ON_PANIC, CTL_EOL);
749#ifdef DIAGNOSTIC
750	sysctl_createv(clog, 0, NULL, NULL,
751		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
752		       CTLTYPE_INT, "panic_now",
753		       SYSCTL_DESCR("Trigger a panic"),
754		       sysctl_kern_trigger_panic, 0, NULL, 0,
755		       CTL_KERN, CTL_CREATE, CTL_EOL);
756#endif
757	sysctl_createv(clog, 0, NULL, NULL,
758		       CTLFLAG_PERMANENT,
759		       CTLTYPE_INT, "root_partition",
760		       SYSCTL_DESCR("Root partition on the root device"),
761		       sysctl_kern_root_partition, 0, NULL, 0,
762		       CTL_KERN, KERN_ROOT_PARTITION, CTL_EOL);
763	sysctl_createv(clog, 0, NULL, NULL,
764		       CTLFLAG_PERMANENT,
765		       CTLTYPE_STRUCT, "drivers",
766		       SYSCTL_DESCR("List of all drivers with block and "
767				    "character device numbers"),
768		       sysctl_kern_drivers, 0, NULL, 0,
769		       CTL_KERN, KERN_DRIVERS, CTL_EOL);
770	sysctl_createv(clog, 0, NULL, NULL,
771		       CTLFLAG_PERMANENT,
772		       CTLTYPE_STRUCT, "file2",
773		       SYSCTL_DESCR("System open file table"),
774		       sysctl_kern_file2, 0, NULL, 0,
775		       CTL_KERN, KERN_FILE2, CTL_EOL);
776	sysctl_createv(clog, 0, NULL, NULL,
777		       CTLFLAG_PERMANENT,
778		       CTLTYPE_STRUCT, "cp_id",
779		       SYSCTL_DESCR("Mapping of CPU number to CPU id"),
780		       sysctl_kern_cpid, 0, NULL, 0,
781		       CTL_KERN, KERN_CP_ID, CTL_EOL);
782	sysctl_createv(clog, 0, NULL, &rnode,
783		       CTLFLAG_PERMANENT,
784		       CTLTYPE_NODE, "coredump",
785		       SYSCTL_DESCR("Coredump settings."),
786		       NULL, 0, NULL, 0,
787		       CTL_KERN, CTL_CREATE, CTL_EOL);
788	sysctl_createv(clog, 0, &rnode, &rnode,
789		       CTLFLAG_PERMANENT,
790		       CTLTYPE_NODE, "setid",
791		       SYSCTL_DESCR("Set-id processes' coredump settings."),
792		       NULL, 0, NULL, 0,
793		       CTL_CREATE, CTL_EOL);
794	sysctl_createv(clog, 0, &rnode, NULL,
795		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
796		       CTLTYPE_INT, "dump",
797		       SYSCTL_DESCR("Allow set-id processes to dump core."),
798		       sysctl_security_setidcore, 0, &security_setidcore_dump,
799		       sizeof(security_setidcore_dump),
800		       CTL_CREATE, CTL_EOL);
801	sysctl_createv(clog, 0, &rnode, NULL,
802		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
803		       CTLTYPE_STRING, "path",
804		       SYSCTL_DESCR("Path pattern for set-id coredumps."),
805		       sysctl_security_setidcorename, 0,
806		       &security_setidcore_path,
807		       sizeof(security_setidcore_path),
808		       CTL_CREATE, CTL_EOL);
809	sysctl_createv(clog, 0, &rnode, NULL,
810		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
811		       CTLTYPE_INT, "owner",
812		       SYSCTL_DESCR("Owner id for set-id processes' cores."),
813		       sysctl_security_setidcore, 0, &security_setidcore_owner,
814		       0,
815		       CTL_CREATE, CTL_EOL);
816	sysctl_createv(clog, 0, &rnode, NULL,
817		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
818		       CTLTYPE_INT, "group",
819		       SYSCTL_DESCR("Group id for set-id processes' cores."),
820		       sysctl_security_setidcore, 0, &security_setidcore_group,
821		       0,
822		       CTL_CREATE, CTL_EOL);
823	sysctl_createv(clog, 0, &rnode, NULL,
824		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
825		       CTLTYPE_INT, "mode",
826		       SYSCTL_DESCR("Mode for set-id processes' cores."),
827		       sysctl_security_setidcore, 0, &security_setidcore_mode,
828		       0,
829		       CTL_CREATE, CTL_EOL);
830	sysctl_createv(clog, 0, NULL, NULL,
831		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
832		       CTLTYPE_INT, "no_sa_support",
833		       SYSCTL_DESCR("0 if the kernel supports SA, otherwise it doesn't"),
834		       NULL, 1, NULL, 0,
835		       CTL_KERN, CTL_CREATE, CTL_EOL);
836}
837
838SYSCTL_SETUP(sysctl_kern_proc_setup,
839	     "sysctl kern.proc/proc2/proc_args subtree setup")
840{
841
842	sysctl_createv(clog, 0, NULL, NULL,
843		       CTLFLAG_PERMANENT,
844		       CTLTYPE_NODE, "kern", NULL,
845		       NULL, 0, NULL, 0,
846		       CTL_KERN, CTL_EOL);
847
848	sysctl_createv(clog, 0, NULL, NULL,
849		       CTLFLAG_PERMANENT,
850		       CTLTYPE_NODE, "proc",
851		       SYSCTL_DESCR("System-wide process information"),
852		       sysctl_doeproc, 0, NULL, 0,
853		       CTL_KERN, KERN_PROC, CTL_EOL);
854	sysctl_createv(clog, 0, NULL, NULL,
855		       CTLFLAG_PERMANENT,
856		       CTLTYPE_NODE, "proc2",
857		       SYSCTL_DESCR("Machine-independent process information"),
858		       sysctl_doeproc, 0, NULL, 0,
859		       CTL_KERN, KERN_PROC2, CTL_EOL);
860	sysctl_createv(clog, 0, NULL, NULL,
861		       CTLFLAG_PERMANENT,
862		       CTLTYPE_NODE, "proc_args",
863		       SYSCTL_DESCR("Process argument information"),
864		       sysctl_kern_proc_args, 0, NULL, 0,
865		       CTL_KERN, KERN_PROC_ARGS, CTL_EOL);
866
867	/*
868	  "nodes" under these:
869
870	  KERN_PROC_ALL
871	  KERN_PROC_PID pid
872	  KERN_PROC_PGRP pgrp
873	  KERN_PROC_SESSION sess
874	  KERN_PROC_TTY tty
875	  KERN_PROC_UID uid
876	  KERN_PROC_RUID uid
877	  KERN_PROC_GID gid
878	  KERN_PROC_RGID gid
879
880	  all in all, probably not worth the effort...
881	*/
882}
883
884SYSCTL_SETUP(sysctl_hw_setup, "sysctl hw subtree setup")
885{
886	u_int u;
887	u_quad_t q;
888
889	sysctl_createv(clog, 0, NULL, NULL,
890		       CTLFLAG_PERMANENT,
891		       CTLTYPE_NODE, "hw", NULL,
892		       NULL, 0, NULL, 0,
893		       CTL_HW, CTL_EOL);
894
895	sysctl_createv(clog, 0, NULL, NULL,
896		       CTLFLAG_PERMANENT,
897		       CTLTYPE_STRING, "machine",
898		       SYSCTL_DESCR("Machine class"),
899		       NULL, 0, machine, 0,
900		       CTL_HW, HW_MACHINE, CTL_EOL);
901	sysctl_createv(clog, 0, NULL, NULL,
902		       CTLFLAG_PERMANENT,
903		       CTLTYPE_STRING, "model",
904		       SYSCTL_DESCR("Machine model"),
905		       NULL, 0, cpu_model, 0,
906		       CTL_HW, HW_MODEL, CTL_EOL);
907	sysctl_createv(clog, 0, NULL, NULL,
908		       CTLFLAG_PERMANENT,
909		       CTLTYPE_INT, "ncpu",
910		       SYSCTL_DESCR("Number of CPUs configured"),
911		       NULL, 0, &ncpu, 0,
912		       CTL_HW, HW_NCPU, CTL_EOL);
913	sysctl_createv(clog, 0, NULL, NULL,
914		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
915		       CTLTYPE_INT, "byteorder",
916		       SYSCTL_DESCR("System byte order"),
917		       NULL, BYTE_ORDER, NULL, 0,
918		       CTL_HW, HW_BYTEORDER, CTL_EOL);
919	u = ((u_int)physmem > (UINT_MAX / PAGE_SIZE)) ?
920		UINT_MAX : physmem * PAGE_SIZE;
921	sysctl_createv(clog, 0, NULL, NULL,
922		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
923		       CTLTYPE_INT, "physmem",
924		       SYSCTL_DESCR("Bytes of physical memory"),
925		       NULL, u, NULL, 0,
926		       CTL_HW, HW_PHYSMEM, CTL_EOL);
927	sysctl_createv(clog, 0, NULL, NULL,
928		       CTLFLAG_PERMANENT,
929		       CTLTYPE_INT, "usermem",
930		       SYSCTL_DESCR("Bytes of non-kernel memory"),
931		       sysctl_hw_usermem, 0, NULL, 0,
932		       CTL_HW, HW_USERMEM, CTL_EOL);
933	sysctl_createv(clog, 0, NULL, NULL,
934		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
935		       CTLTYPE_INT, "pagesize",
936		       SYSCTL_DESCR("Software page size"),
937		       NULL, PAGE_SIZE, NULL, 0,
938		       CTL_HW, HW_PAGESIZE, CTL_EOL);
939	sysctl_createv(clog, 0, NULL, NULL,
940		       CTLFLAG_PERMANENT,
941		       CTLTYPE_STRING, "machine_arch",
942		       SYSCTL_DESCR("Machine CPU class"),
943		       NULL, 0, machine_arch, 0,
944		       CTL_HW, HW_MACHINE_ARCH, CTL_EOL);
945	sysctl_createv(clog, 0, NULL, NULL,
946		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
947		       CTLTYPE_INT, "alignbytes",
948		       SYSCTL_DESCR("Alignment constraint for all possible "
949				    "data types"),
950		       NULL, ALIGNBYTES, NULL, 0,
951		       CTL_HW, HW_ALIGNBYTES, CTL_EOL);
952	sysctl_createv(clog, 0, NULL, NULL,
953		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE|CTLFLAG_HEX,
954		       CTLTYPE_STRING, "cnmagic",
955		       SYSCTL_DESCR("Console magic key sequence"),
956		       sysctl_hw_cnmagic, 0, NULL, CNS_LEN,
957		       CTL_HW, HW_CNMAGIC, CTL_EOL);
958	q = (u_quad_t)physmem * PAGE_SIZE;
959	sysctl_createv(clog, 0, NULL, NULL,
960		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
961		       CTLTYPE_QUAD, "physmem64",
962		       SYSCTL_DESCR("Bytes of physical memory"),
963		       NULL, q, NULL, 0,
964		       CTL_HW, HW_PHYSMEM64, CTL_EOL);
965	sysctl_createv(clog, 0, NULL, NULL,
966		       CTLFLAG_PERMANENT,
967		       CTLTYPE_QUAD, "usermem64",
968		       SYSCTL_DESCR("Bytes of non-kernel memory"),
969		       sysctl_hw_usermem, 0, NULL, 0,
970		       CTL_HW, HW_USERMEM64, CTL_EOL);
971	sysctl_createv(clog, 0, NULL, NULL,
972		       CTLFLAG_PERMANENT,
973		       CTLTYPE_INT, "ncpuonline",
974		       SYSCTL_DESCR("Number of CPUs online"),
975		       NULL, 0, &ncpuonline, 0,
976		       CTL_HW, HW_NCPUONLINE, CTL_EOL);
977}
978
979#ifdef DEBUG
980/*
981 * Debugging related system variables.
982 */
983struct ctldebug /* debug0, */ /* debug1, */ debug2, debug3, debug4;
984struct ctldebug debug5, debug6, debug7, debug8, debug9;
985struct ctldebug debug10, debug11, debug12, debug13, debug14;
986struct ctldebug debug15, debug16, debug17, debug18, debug19;
987static struct ctldebug *debugvars[CTL_DEBUG_MAXID] = {
988	&debug0, &debug1, &debug2, &debug3, &debug4,
989	&debug5, &debug6, &debug7, &debug8, &debug9,
990	&debug10, &debug11, &debug12, &debug13, &debug14,
991	&debug15, &debug16, &debug17, &debug18, &debug19,
992};
993
994/*
995 * this setup routine is a replacement for debug_sysctl()
996 *
997 * note that it creates several nodes per defined debug variable
998 */
999SYSCTL_SETUP(sysctl_debug_setup, "sysctl debug subtree setup")
1000{
1001	struct ctldebug *cdp;
1002	char nodename[20];
1003	int i;
1004
1005	/*
1006	 * two ways here:
1007	 *
1008	 * the "old" way (debug.name -> value) which was emulated by
1009	 * the sysctl(8) binary
1010	 *
1011	 * the new way, which the sysctl(8) binary was actually using
1012
1013	 node	debug
1014	 node	debug.0
1015	 string debug.0.name
1016	 int	debug.0.value
1017	 int	debug.name
1018
1019	 */
1020
1021	sysctl_createv(clog, 0, NULL, NULL,
1022		       CTLFLAG_PERMANENT,
1023		       CTLTYPE_NODE, "debug", NULL,
1024		       NULL, 0, NULL, 0,
1025		       CTL_DEBUG, CTL_EOL);
1026
1027	for (i = 0; i < CTL_DEBUG_MAXID; i++) {
1028		cdp = debugvars[i];
1029		if (cdp->debugname == NULL || cdp->debugvar == NULL)
1030			continue;
1031
1032		snprintf(nodename, sizeof(nodename), "debug%d", i);
1033		sysctl_createv(clog, 0, NULL, NULL,
1034			       CTLFLAG_PERMANENT|CTLFLAG_HIDDEN,
1035			       CTLTYPE_NODE, nodename, NULL,
1036			       NULL, 0, NULL, 0,
1037			       CTL_DEBUG, i, CTL_EOL);
1038		sysctl_createv(clog, 0, NULL, NULL,
1039			       CTLFLAG_PERMANENT|CTLFLAG_HIDDEN,
1040			       CTLTYPE_STRING, "name", NULL,
1041			       /*XXXUNCONST*/
1042			       NULL, 0, __UNCONST(cdp->debugname), 0,
1043			       CTL_DEBUG, i, CTL_DEBUG_NAME, CTL_EOL);
1044		sysctl_createv(clog, 0, NULL, NULL,
1045			       CTLFLAG_PERMANENT|CTLFLAG_HIDDEN,
1046			       CTLTYPE_INT, "value", NULL,
1047			       NULL, 0, cdp->debugvar, 0,
1048			       CTL_DEBUG, i, CTL_DEBUG_VALUE, CTL_EOL);
1049		sysctl_createv(clog, 0, NULL, NULL,
1050			       CTLFLAG_PERMANENT,
1051			       CTLTYPE_INT, cdp->debugname, NULL,
1052			       NULL, 0, cdp->debugvar, 0,
1053			       CTL_DEBUG, CTL_CREATE, CTL_EOL);
1054	}
1055}
1056#endif /* DEBUG */
1057
1058/*
1059 * ********************************************************************
1060 * section 2: private node-specific helper routines.
1061 * ********************************************************************
1062 */
1063
1064#ifdef DIAGNOSTIC
1065static int
1066sysctl_kern_trigger_panic(SYSCTLFN_ARGS)
1067{
1068	int newtrig, error;
1069	struct sysctlnode node;
1070
1071	newtrig = 0;
1072	node = *rnode;
1073	node.sysctl_data = &newtrig;
1074	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1075	if (error || newp == NULL)
1076		return (error);
1077
1078	if (newtrig != 0)
1079		panic("Panic triggered");
1080
1081	return (error);
1082}
1083#endif
1084
1085/*
1086 * sysctl helper routine for kern.maxvnodes.  Drain vnodes if
1087 * new value is lower than desiredvnodes and then calls reinit
1088 * routines that needs to adjust to the new value.
1089 */
1090static int
1091sysctl_kern_maxvnodes(SYSCTLFN_ARGS)
1092{
1093	int error, new_vnodes, old_vnodes;
1094	struct sysctlnode node;
1095
1096	new_vnodes = desiredvnodes;
1097	node = *rnode;
1098	node.sysctl_data = &new_vnodes;
1099	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1100	if (error || newp == NULL)
1101		return (error);
1102
1103	old_vnodes = desiredvnodes;
1104	desiredvnodes = new_vnodes;
1105	if (new_vnodes < old_vnodes) {
1106		error = vfs_drainvnodes(new_vnodes, l);
1107		if (error) {
1108			desiredvnodes = old_vnodes;
1109			return (error);
1110		}
1111	}
1112	vfs_reinit();
1113	nchreinit();
1114
1115	return (0);
1116}
1117
1118/*
1119 * sysctl helper routine for rtc_offset - set time after changes
1120 */
1121static int
1122sysctl_kern_rtc_offset(SYSCTLFN_ARGS)
1123{
1124	struct timespec ts, delta;
1125	int error, new_rtc_offset;
1126	struct sysctlnode node;
1127
1128	new_rtc_offset = rtc_offset;
1129	node = *rnode;
1130	node.sysctl_data = &new_rtc_offset;
1131	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1132	if (error || newp == NULL)
1133		return (error);
1134
1135	if (kauth_authorize_system(l->l_cred, KAUTH_SYSTEM_TIME,
1136	    KAUTH_REQ_SYSTEM_TIME_RTCOFFSET,
1137	    KAUTH_ARG(new_rtc_offset), NULL, NULL))
1138		return (EPERM);
1139	if (rtc_offset == new_rtc_offset)
1140		return (0);
1141
1142	/* if we change the offset, adjust the time */
1143	nanotime(&ts);
1144	delta.tv_sec = 60 * (new_rtc_offset - rtc_offset);
1145	delta.tv_nsec = 0;
1146	timespecadd(&ts, &delta, &ts);
1147	rtc_offset = new_rtc_offset;
1148	return (settime(l->l_proc, &ts));
1149}
1150
1151/*
1152 * sysctl helper routine for kern.maxproc. Ensures that the new
1153 * values are not too low or too high.
1154 */
1155static int
1156sysctl_kern_maxproc(SYSCTLFN_ARGS)
1157{
1158	int error, nmaxproc;
1159	struct sysctlnode node;
1160
1161	nmaxproc = maxproc;
1162	node = *rnode;
1163	node.sysctl_data = &nmaxproc;
1164	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1165	if (error || newp == NULL)
1166		return (error);
1167
1168	if (nmaxproc < 0 || nmaxproc >= PID_MAX)
1169		return (EINVAL);
1170#ifdef __HAVE_CPU_MAXPROC
1171	if (nmaxproc > cpu_maxproc())
1172		return (EINVAL);
1173#endif
1174	maxproc = nmaxproc;
1175
1176	return (0);
1177}
1178
1179/*
1180 * sysctl helper function for kern.hostid. The hostid is a long, but
1181 * we export it as an int, so we need to give it a little help.
1182 */
1183static int
1184sysctl_kern_hostid(SYSCTLFN_ARGS)
1185{
1186	int error, inthostid;
1187	struct sysctlnode node;
1188
1189	inthostid = hostid;  /* XXX assumes sizeof int <= sizeof long */
1190	node = *rnode;
1191	node.sysctl_data = &inthostid;
1192	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1193	if (error || newp == NULL)
1194		return (error);
1195
1196	hostid = (unsigned)inthostid;
1197
1198	return (0);
1199}
1200
1201/*
1202 * sysctl helper function for kern.hostname and kern.domainnname.
1203 * resets the relevant recorded length when the underlying name is
1204 * changed.
1205 */
1206static int
1207sysctl_setlen(SYSCTLFN_ARGS)
1208{
1209	int error;
1210
1211	error = sysctl_lookup(SYSCTLFN_CALL(rnode));
1212	if (error || newp == NULL)
1213		return (error);
1214
1215	switch (rnode->sysctl_num) {
1216	case KERN_HOSTNAME:
1217		hostnamelen = strlen((const char*)rnode->sysctl_data);
1218		break;
1219	case KERN_DOMAINNAME:
1220		domainnamelen = strlen((const char*)rnode->sysctl_data);
1221		break;
1222	}
1223
1224	return (0);
1225}
1226
1227/*
1228 * sysctl helper routine for kern.clockrate. Assembles a struct on
1229 * the fly to be returned to the caller.
1230 */
1231static int
1232sysctl_kern_clockrate(SYSCTLFN_ARGS)
1233{
1234	struct clockinfo clkinfo;
1235	struct sysctlnode node;
1236
1237	clkinfo.tick = tick;
1238	clkinfo.tickadj = tickadj;
1239	clkinfo.hz = hz;
1240	clkinfo.profhz = profhz;
1241	clkinfo.stathz = stathz ? stathz : hz;
1242
1243	node = *rnode;
1244	node.sysctl_data = &clkinfo;
1245	return (sysctl_lookup(SYSCTLFN_CALL(&node)));
1246}
1247
1248
1249/*
1250 * sysctl helper routine for kern.file pseudo-subtree.
1251 */
1252static int
1253sysctl_kern_file(SYSCTLFN_ARGS)
1254{
1255	int error;
1256	size_t buflen;
1257	struct file *fp, *dp, *np, fbuf;
1258	char *start, *where;
1259
1260	start = where = oldp;
1261	buflen = *oldlenp;
1262	dp = NULL;
1263
1264	if (where == NULL) {
1265		/*
1266		 * overestimate by 10 files
1267		 */
1268		*oldlenp = sizeof(filehead) + (nfiles + 10) *
1269		    sizeof(struct file);
1270		return (0);
1271	}
1272
1273	/*
1274	 * first dcopyout filehead
1275	 */
1276	if (buflen < sizeof(filehead)) {
1277		*oldlenp = 0;
1278		return (0);
1279	}
1280	sysctl_unlock();
1281	error = dcopyout(l, &filehead, where, sizeof(filehead));
1282	if (error) {
1283	 	sysctl_relock();
1284		return error;
1285	}
1286	buflen -= sizeof(filehead);
1287	where += sizeof(filehead);
1288
1289	/*
1290	 * allocate dummy file descriptor to make position in list
1291	 */
1292	if ((dp = fgetdummy()) == NULL) {
1293	 	sysctl_relock();
1294		return ENOMEM;
1295	}
1296
1297	/*
1298	 * followed by an array of file structures
1299	 */
1300	mutex_enter(&filelist_lock);
1301	for (fp = LIST_FIRST(&filehead); fp != NULL; fp = np) {
1302	    	np = LIST_NEXT(fp, f_list);
1303	    	mutex_enter(&fp->f_lock);
1304	    	if (fp->f_count == 0) {
1305		    	mutex_exit(&fp->f_lock);
1306	    		continue;
1307		}
1308		/*
1309		 * XXX Need to prevent that from being an alternative way
1310		 * XXX to getting process information.
1311		 */
1312		if (kauth_authorize_generic(l->l_cred,
1313		    KAUTH_GENERIC_CANSEE, fp->f_cred) != 0) {
1314		    	mutex_exit(&fp->f_lock);
1315			continue;
1316		}
1317		if (buflen < sizeof(struct file)) {
1318			*oldlenp = where - start;
1319		    	mutex_exit(&fp->f_lock);
1320			error = ENOMEM;
1321			break;
1322		}
1323		memcpy(&fbuf, fp, sizeof(fbuf));
1324		LIST_INSERT_AFTER(fp, dp, f_list);
1325	    	mutex_exit(&fp->f_lock);
1326		mutex_exit(&filelist_lock);
1327		error = dcopyout(l, &fbuf, where, sizeof(fbuf));
1328		if (error) {
1329			mutex_enter(&filelist_lock);
1330			LIST_REMOVE(dp, f_list);
1331			break;
1332		}
1333		buflen -= sizeof(struct file);
1334		where += sizeof(struct file);
1335		mutex_enter(&filelist_lock);
1336		np = LIST_NEXT(dp, f_list);
1337		LIST_REMOVE(dp, f_list);
1338	}
1339	mutex_exit(&filelist_lock);
1340	*oldlenp = where - start;
1341 	if (dp != NULL)
1342		fputdummy(dp);
1343 	sysctl_relock();
1344	return (error);
1345}
1346
1347/*
1348 * sysctl helper routine for kern.msgbufsize and kern.msgbuf. For the
1349 * former it merely checks the message buffer is set up. For the latter,
1350 * it also copies out the data if necessary.
1351 */
1352static int
1353sysctl_msgbuf(SYSCTLFN_ARGS)
1354{
1355	char *where = oldp;
1356	size_t len, maxlen;
1357	long beg, end;
1358	extern kmutex_t log_lock;
1359	int error;
1360
1361	if (!msgbufenabled || msgbufp->msg_magic != MSG_MAGIC) {
1362		msgbufenabled = 0;
1363		return (ENXIO);
1364	}
1365
1366	switch (rnode->sysctl_num) {
1367	case KERN_MSGBUFSIZE: {
1368		struct sysctlnode node = *rnode;
1369		int msg_bufs = (int)msgbufp->msg_bufs;
1370		node.sysctl_data = &msg_bufs;
1371		return (sysctl_lookup(SYSCTLFN_CALL(&node)));
1372	}
1373	case KERN_MSGBUF:
1374		break;
1375	default:
1376		return (EOPNOTSUPP);
1377	}
1378
1379	if (newp != NULL)
1380		return (EPERM);
1381
1382	if (oldp == NULL) {
1383		/* always return full buffer size */
1384		*oldlenp = msgbufp->msg_bufs;
1385		return (0);
1386	}
1387
1388	sysctl_unlock();
1389
1390	/*
1391	 * First, copy from the write pointer to the end of
1392	 * message buffer.
1393	 */
1394	error = 0;
1395	mutex_spin_enter(&log_lock);
1396	maxlen = MIN(msgbufp->msg_bufs, *oldlenp);
1397	beg = msgbufp->msg_bufx;
1398	end = msgbufp->msg_bufs;
1399	mutex_spin_exit(&log_lock);
1400
1401	while (maxlen > 0) {
1402		len = MIN(end - beg, maxlen);
1403		if (len == 0)
1404			break;
1405		/* XXX unlocked, but hardly matters. */
1406		error = dcopyout(l, &msgbufp->msg_bufc[beg], where, len);
1407		if (error)
1408			break;
1409		where += len;
1410		maxlen -= len;
1411
1412		/*
1413		 * ... then, copy from the beginning of message buffer to
1414		 * the write pointer.
1415		 */
1416		beg = 0;
1417		end = msgbufp->msg_bufx;
1418	}
1419
1420	sysctl_relock();
1421	return (error);
1422}
1423
1424/*
1425 * sysctl helper routine for kern.defcorename. In the case of a new
1426 * string being assigned, check that it's not a zero-length string.
1427 * (XXX the check in -current doesn't work, but do we really care?)
1428 */
1429static int
1430sysctl_kern_defcorename(SYSCTLFN_ARGS)
1431{
1432	int error;
1433	char *newcorename;
1434	struct sysctlnode node;
1435
1436	newcorename = PNBUF_GET();
1437	node = *rnode;
1438	node.sysctl_data = &newcorename[0];
1439	memcpy(node.sysctl_data, rnode->sysctl_data, MAXPATHLEN);
1440	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1441	if (error || newp == NULL) {
1442		goto done;
1443	}
1444
1445	/*
1446	 * when sysctl_lookup() deals with a string, it's guaranteed
1447	 * to come back nul terminated. So there.  :)
1448	 */
1449	if (strlen(newcorename) == 0) {
1450		error = EINVAL;
1451	} else {
1452		memcpy(rnode->sysctl_data, node.sysctl_data, MAXPATHLEN);
1453		error = 0;
1454	}
1455done:
1456	PNBUF_PUT(newcorename);
1457	return error;
1458}
1459
1460/*
1461 * sysctl helper routine for kern.cp_time node. Adds up cpu time
1462 * across all cpus.
1463 */
1464static int
1465sysctl_kern_cptime(SYSCTLFN_ARGS)
1466{
1467	struct sysctlnode node = *rnode;
1468	uint64_t *cp_time = NULL;
1469	int error, n = ncpu, i;
1470	struct cpu_info *ci;
1471	CPU_INFO_ITERATOR cii;
1472
1473	/*
1474	 * if you specifically pass a buffer that is the size of the
1475	 * sum, or if you are probing for the size, you get the "sum"
1476	 * of cp_time (and the size thereof) across all processors.
1477	 *
1478	 * alternately, you can pass an additional mib number and get
1479	 * cp_time for that particular processor.
1480	 */
1481	switch (namelen) {
1482	case 0:
1483		if (*oldlenp == sizeof(uint64_t) * CPUSTATES || oldp == NULL) {
1484			node.sysctl_size = sizeof(uint64_t) * CPUSTATES;
1485			n = -1; /* SUM */
1486		}
1487		else {
1488			node.sysctl_size = n * sizeof(uint64_t) * CPUSTATES;
1489			n = -2; /* ALL */
1490		}
1491		break;
1492	case 1:
1493		if (name[0] < 0 || name[0] >= n)
1494			return (ENOENT); /* ENOSUCHPROCESSOR */
1495		node.sysctl_size = sizeof(uint64_t) * CPUSTATES;
1496		n = name[0];
1497		/*
1498		 * adjust these so that sysctl_lookup() will be happy
1499		 */
1500		name++;
1501		namelen--;
1502		break;
1503	default:
1504		return (EINVAL);
1505	}
1506
1507	cp_time = kmem_alloc(node.sysctl_size, KM_SLEEP);
1508	if (cp_time == NULL)
1509		return (ENOMEM);
1510	node.sysctl_data = cp_time;
1511	memset(cp_time, 0, node.sysctl_size);
1512
1513	for (CPU_INFO_FOREACH(cii, ci)) {
1514		if (n <= 0) {
1515			for (i = 0; i < CPUSTATES; i++) {
1516				cp_time[i] += ci->ci_schedstate.spc_cp_time[i];
1517			}
1518		}
1519		/*
1520		 * if a specific processor was requested and we just
1521		 * did it, we're done here
1522		 */
1523		if (n == 0)
1524			break;
1525		/*
1526		 * if doing "all", skip to next cp_time set for next processor
1527		 */
1528		if (n == -2)
1529			cp_time += CPUSTATES;
1530		/*
1531		 * if we're doing a specific processor, we're one
1532		 * processor closer
1533		 */
1534		if (n > 0)
1535			n--;
1536	}
1537
1538	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1539	kmem_free(node.sysctl_data, node.sysctl_size);
1540	return (error);
1541}
1542
1543#if NPTY > 0
1544/*
1545 * sysctl helper routine for kern.maxptys. Ensures that any new value
1546 * is acceptable to the pty subsystem.
1547 */
1548static int
1549sysctl_kern_maxptys(SYSCTLFN_ARGS)
1550{
1551	int pty_maxptys(int, int);		/* defined in kern/tty_pty.c */
1552	int error, xmax;
1553	struct sysctlnode node;
1554
1555	/* get current value of maxptys */
1556	xmax = pty_maxptys(0, 0);
1557
1558	node = *rnode;
1559	node.sysctl_data = &xmax;
1560	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1561	if (error || newp == NULL)
1562		return (error);
1563
1564	if (xmax != pty_maxptys(xmax, 1))
1565		return (EINVAL);
1566
1567	return (0);
1568}
1569#endif /* NPTY > 0 */
1570
1571/*
1572 * sysctl helper routine for kern.sbmax. Basically just ensures that
1573 * any new value is not too small.
1574 */
1575static int
1576sysctl_kern_sbmax(SYSCTLFN_ARGS)
1577{
1578	int error, new_sbmax;
1579	struct sysctlnode node;
1580
1581	new_sbmax = sb_max;
1582	node = *rnode;
1583	node.sysctl_data = &new_sbmax;
1584	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1585	if (error || newp == NULL)
1586		return (error);
1587
1588	KERNEL_LOCK(1, NULL);
1589	error = sb_max_set(new_sbmax);
1590	KERNEL_UNLOCK_ONE(NULL);
1591
1592	return (error);
1593}
1594
1595/*
1596 * sysctl helper routine for kern.urandom node. Picks a random number
1597 * for you.
1598 */
1599static int
1600sysctl_kern_urnd(SYSCTLFN_ARGS)
1601{
1602#if NRND > 0
1603	int v, rv;
1604
1605	KERNEL_LOCK(1, NULL);
1606	rv = rnd_extract_data(&v, sizeof(v), RND_EXTRACT_ANY);
1607	KERNEL_UNLOCK_ONE(NULL);
1608	if (rv == sizeof(v)) {
1609		struct sysctlnode node = *rnode;
1610		node.sysctl_data = &v;
1611		return (sysctl_lookup(SYSCTLFN_CALL(&node)));
1612	}
1613	else
1614		return (EIO);	/*XXX*/
1615#else
1616	return (EOPNOTSUPP);
1617#endif
1618}
1619
1620/*
1621 * sysctl helper routine for kern.arandom node. Picks a random number
1622 * for you.
1623 */
1624static int
1625sysctl_kern_arnd(SYSCTLFN_ARGS)
1626{
1627#if NRND > 0
1628	int error;
1629	void *v;
1630	struct sysctlnode node = *rnode;
1631
1632	if (*oldlenp == 0)
1633		return 0;
1634	if (*oldlenp > 8192)
1635		return E2BIG;
1636
1637	v = kmem_alloc(*oldlenp, KM_SLEEP);
1638	arc4randbytes(v, *oldlenp);
1639	node.sysctl_data = v;
1640	node.sysctl_size = *oldlenp;
1641	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1642	kmem_free(v, *oldlenp);
1643	return error;
1644#else
1645	return (EOPNOTSUPP);
1646#endif
1647}
1648/*
1649 * sysctl helper routine to do kern.lwp.* work.
1650 */
1651static int
1652sysctl_kern_lwp(SYSCTLFN_ARGS)
1653{
1654	struct kinfo_lwp klwp;
1655	struct proc *p;
1656	struct lwp *l2, *l3;
1657	char *where, *dp;
1658	int pid, elem_size, elem_count;
1659	int buflen, needed, error;
1660	bool gotit;
1661
1662	if (namelen == 1 && name[0] == CTL_QUERY)
1663		return (sysctl_query(SYSCTLFN_CALL(rnode)));
1664
1665	dp = where = oldp;
1666	buflen = where != NULL ? *oldlenp : 0;
1667	error = needed = 0;
1668
1669	if (newp != NULL || namelen != 3)
1670		return (EINVAL);
1671	pid = name[0];
1672	elem_size = name[1];
1673	elem_count = name[2];
1674
1675	sysctl_unlock();
1676	if (pid == -1) {
1677		mutex_enter(proc_lock);
1678		LIST_FOREACH(p, &allproc, p_list) {
1679			/* Grab a hold on the process. */
1680			if (!rw_tryenter(&p->p_reflock, RW_READER)) {
1681				continue;
1682			}
1683			mutex_exit(proc_lock);
1684
1685			mutex_enter(p->p_lock);
1686			LIST_FOREACH(l2, &p->p_lwps, l_sibling) {
1687				if (buflen >= elem_size && elem_count > 0) {
1688					lwp_lock(l2);
1689					fill_lwp(l2, &klwp);
1690					lwp_unlock(l2);
1691					mutex_exit(p->p_lock);
1692
1693					/*
1694					 * Copy out elem_size, but not
1695					 * larger than the size of a
1696					 * struct kinfo_proc2.
1697					 */
1698					error = dcopyout(l, &klwp, dp,
1699					    min(sizeof(klwp), elem_size));
1700					if (error) {
1701						rw_exit(&p->p_reflock);
1702						goto cleanup;
1703					}
1704					mutex_enter(p->p_lock);
1705					LIST_FOREACH(l3, &p->p_lwps,
1706					    l_sibling) {
1707						if (l2 == l3)
1708							break;
1709					}
1710					if (l3 == NULL) {
1711						mutex_exit(p->p_lock);
1712						rw_exit(&p->p_reflock);
1713						error = EAGAIN;
1714						goto cleanup;
1715					}
1716					dp += elem_size;
1717					buflen -= elem_size;
1718					elem_count--;
1719				}
1720				needed += elem_size;
1721			}
1722			mutex_exit(p->p_lock);
1723
1724			/* Drop reference to process. */
1725			mutex_enter(proc_lock);
1726			rw_exit(&p->p_reflock);
1727		}
1728		mutex_exit(proc_lock);
1729	} else {
1730		mutex_enter(proc_lock);
1731		p = p_find(pid, PFIND_LOCKED);
1732		if (p == NULL) {
1733			error = ESRCH;
1734			mutex_exit(proc_lock);
1735			goto cleanup;
1736		}
1737		/* Grab a hold on the process. */
1738		gotit = rw_tryenter(&p->p_reflock, RW_READER);
1739		mutex_exit(proc_lock);
1740		if (!gotit) {
1741			error = ESRCH;
1742			goto cleanup;
1743		}
1744
1745		mutex_enter(p->p_lock);
1746		LIST_FOREACH(l2, &p->p_lwps, l_sibling) {
1747			if (buflen >= elem_size && elem_count > 0) {
1748				lwp_lock(l2);
1749				fill_lwp(l2, &klwp);
1750				lwp_unlock(l2);
1751				mutex_exit(p->p_lock);
1752				/*
1753				 * Copy out elem_size, but not larger than
1754				 * the size of a struct kinfo_proc2.
1755				 */
1756				error = dcopyout(l, &klwp, dp,
1757				    min(sizeof(klwp), elem_size));
1758				if (error) {
1759					rw_exit(&p->p_reflock);
1760					goto cleanup;
1761				}
1762				mutex_enter(p->p_lock);
1763				LIST_FOREACH(l3, &p->p_lwps, l_sibling) {
1764					if (l2 == l3)
1765						break;
1766				}
1767				if (l3 == NULL) {
1768					mutex_exit(p->p_lock);
1769					rw_exit(&p->p_reflock);
1770					error = EAGAIN;
1771					goto cleanup;
1772				}
1773				dp += elem_size;
1774				buflen -= elem_size;
1775				elem_count--;
1776			}
1777			needed += elem_size;
1778		}
1779		mutex_exit(p->p_lock);
1780
1781		/* Drop reference to process. */
1782		rw_exit(&p->p_reflock);
1783	}
1784
1785	if (where != NULL) {
1786		*oldlenp = dp - where;
1787		if (needed > *oldlenp) {
1788			sysctl_relock();
1789			return (ENOMEM);
1790		}
1791	} else {
1792		needed += KERN_LWPSLOP;
1793		*oldlenp = needed;
1794	}
1795	error = 0;
1796 cleanup:
1797	sysctl_relock();
1798	return (error);
1799}
1800
1801/*
1802 * sysctl helper routine for kern.forkfsleep node. Ensures that the
1803 * given value is not too large or two small, and is at least one
1804 * timer tick if not zero.
1805 */
1806static int
1807sysctl_kern_forkfsleep(SYSCTLFN_ARGS)
1808{
1809	/* userland sees value in ms, internally is in ticks */
1810	extern int forkfsleep;		/* defined in kern/kern_fork.c */
1811	int error, timo, lsleep;
1812	struct sysctlnode node;
1813
1814	lsleep = forkfsleep * 1000 / hz;
1815	node = *rnode;
1816	node.sysctl_data = &lsleep;
1817	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1818	if (error || newp == NULL)
1819		return (error);
1820
1821	/* refuse negative values, and overly 'long time' */
1822	if (lsleep < 0 || lsleep > MAXSLP * 1000)
1823		return (EINVAL);
1824
1825	timo = mstohz(lsleep);
1826
1827	/* if the interval is >0 ms && <1 tick, use 1 tick */
1828	if (lsleep != 0 && timo == 0)
1829		forkfsleep = 1;
1830	else
1831		forkfsleep = timo;
1832
1833	return (0);
1834}
1835
1836/*
1837 * sysctl helper routine for kern.root_partition
1838 */
1839static int
1840sysctl_kern_root_partition(SYSCTLFN_ARGS)
1841{
1842	int rootpart = DISKPART(rootdev);
1843	struct sysctlnode node = *rnode;
1844
1845	node.sysctl_data = &rootpart;
1846	return (sysctl_lookup(SYSCTLFN_CALL(&node)));
1847}
1848
1849/*
1850 * sysctl helper function for kern.drivers
1851 */
1852static int
1853sysctl_kern_drivers(SYSCTLFN_ARGS)
1854{
1855	int error;
1856	size_t buflen;
1857	struct kinfo_drivers kd;
1858	char *start, *where;
1859	const char *dname;
1860	int i;
1861	extern struct devsw_conv *devsw_conv;
1862	extern int max_devsw_convs;
1863	extern kmutex_t devsw_lock;
1864
1865	if (newp != NULL || namelen != 0)
1866		return (EINVAL);
1867
1868	start = where = oldp;
1869	buflen = *oldlenp;
1870	if (where == NULL) {
1871		*oldlenp = max_devsw_convs * sizeof kd;
1872		return 0;
1873	}
1874
1875	/*
1876	 * An array of kinfo_drivers structures
1877	 */
1878	error = 0;
1879	sysctl_unlock();
1880	mutex_enter(&devsw_lock);
1881	for (i = 0; i < max_devsw_convs; i++) {
1882		dname = devsw_conv[i].d_name;
1883		if (dname == NULL)
1884			continue;
1885		if (buflen < sizeof kd) {
1886			error = ENOMEM;
1887			break;
1888		}
1889		memset(&kd, 0, sizeof(kd));
1890		kd.d_bmajor = devsw_conv[i].d_bmajor;
1891		kd.d_cmajor = devsw_conv[i].d_cmajor;
1892		strlcpy(kd.d_name, dname, sizeof kd.d_name);
1893		mutex_exit(&devsw_lock);
1894		error = dcopyout(l, &kd, where, sizeof kd);
1895		mutex_enter(&devsw_lock);
1896		if (error != 0)
1897			break;
1898		buflen -= sizeof kd;
1899		where += sizeof kd;
1900	}
1901	mutex_exit(&devsw_lock);
1902	sysctl_relock();
1903	*oldlenp = where - start;
1904	return error;
1905}
1906
1907/*
1908 * sysctl helper function for kern.file2
1909 */
1910static int
1911sysctl_kern_file2(SYSCTLFN_ARGS)
1912{
1913	struct proc *p;
1914	struct file *fp, *tp, *np;
1915	struct filedesc *fd;
1916	struct kinfo_file kf;
1917	char *dp;
1918	u_int i, op;
1919	size_t len, needed, elem_size, out_size;
1920	int error, arg, elem_count;
1921	fdfile_t *ff;
1922
1923	if (namelen == 1 && name[0] == CTL_QUERY)
1924		return (sysctl_query(SYSCTLFN_CALL(rnode)));
1925
1926	if (namelen != 4)
1927		return (EINVAL);
1928
1929	error = 0;
1930	dp = oldp;
1931	len = (oldp != NULL) ? *oldlenp : 0;
1932	op = name[0];
1933	arg = name[1];
1934	elem_size = name[2];
1935	elem_count = name[3];
1936	out_size = MIN(sizeof(kf), elem_size);
1937	needed = 0;
1938
1939	if (elem_size < 1 || elem_count < 0)
1940		return (EINVAL);
1941
1942	switch (op) {
1943	case KERN_FILE_BYFILE:
1944		/*
1945		 * doesn't use arg so it must be zero
1946		 */
1947		if (arg != 0)
1948			return (EINVAL);
1949		sysctl_unlock();
1950		/*
1951		 * allocate dummy file descriptor to make position in list
1952		 */
1953		if ((tp = fgetdummy()) == NULL) {
1954		 	sysctl_relock();
1955			return ENOMEM;
1956		}
1957		mutex_enter(&filelist_lock);
1958		for (fp = LIST_FIRST(&filehead); fp != NULL; fp = np) {
1959			np = LIST_NEXT(fp, f_list);
1960			mutex_enter(&fp->f_lock);
1961			if (fp->f_count == 0) {
1962				mutex_exit(&fp->f_lock);
1963				continue;
1964			}
1965			/*
1966			 * XXX Need to prevent that from being an alternative
1967			 * XXX way for getting process information.
1968			 */
1969			if (kauth_authorize_generic(l->l_cred,
1970			    KAUTH_GENERIC_CANSEE, fp->f_cred) != 0) {
1971				mutex_exit(&fp->f_lock);
1972				continue;
1973			}
1974			if (len >= elem_size && elem_count > 0) {
1975				fill_file(&kf, fp, NULL, 0, 0);
1976				LIST_INSERT_AFTER(fp, tp, f_list);
1977				mutex_exit(&fp->f_lock);
1978				mutex_exit(&filelist_lock);
1979				error = dcopyout(l, &kf, dp, out_size);
1980				mutex_enter(&filelist_lock);
1981				np = LIST_NEXT(tp, f_list);
1982				LIST_REMOVE(tp, f_list);
1983				if (error) {
1984					break;
1985				}
1986				dp += elem_size;
1987				len -= elem_size;
1988			} else {
1989				mutex_exit(&fp->f_lock);
1990			}
1991			if (elem_count > 0) {
1992				needed += elem_size;
1993				if (elem_count != INT_MAX)
1994					elem_count--;
1995			}
1996		}
1997		mutex_exit(&filelist_lock);
1998		fputdummy(tp);
1999		sysctl_relock();
2000		break;
2001	case KERN_FILE_BYPID:
2002		if (arg < -1)
2003			/* -1 means all processes */
2004			return (EINVAL);
2005		sysctl_unlock();
2006		mutex_enter(proc_lock);
2007		LIST_FOREACH(p, &allproc, p_list) {
2008			if (p->p_stat == SIDL) {
2009				/* skip embryonic processes */
2010				continue;
2011			}
2012			if (arg > 0 && p->p_pid != arg) {
2013				/* pick only the one we want */
2014				/* XXX want 0 to mean "kernel files" */
2015				continue;
2016			}
2017			mutex_enter(p->p_lock);
2018			error = kauth_authorize_process(l->l_cred,
2019			    KAUTH_PROCESS_CANSEE, p,
2020			    KAUTH_ARG(KAUTH_REQ_PROCESS_CANSEE_OPENFILES),
2021			    NULL, NULL);
2022			mutex_exit(p->p_lock);
2023			if (error != 0) {
2024				continue;
2025			}
2026
2027			/*
2028			 * Grab a hold on the process.
2029			 */
2030			if (!rw_tryenter(&p->p_reflock, RW_READER)) {
2031				continue;
2032			}
2033			mutex_exit(proc_lock);
2034
2035			/* XXX Do we need to check permission per file? */
2036			fd = p->p_fd;
2037			mutex_enter(&fd->fd_lock);
2038			for (i = 0; i < fd->fd_nfiles; i++) {
2039				if ((ff = fd->fd_ofiles[i]) == NULL) {
2040					continue;
2041				}
2042				mutex_enter(&ff->ff_lock);
2043				if ((fp = ff->ff_file) == NULL) {
2044					mutex_exit(&ff->ff_lock);
2045					continue;
2046				}
2047				if (len >= elem_size && elem_count > 0) {
2048					fill_file(&kf, fp, ff, i, p->p_pid);
2049					mutex_exit(&fp->f_lock);
2050					mutex_exit(&ff->ff_lock);
2051					mutex_exit(&fd->fd_lock);
2052					error = dcopyout(l, &kf, dp, out_size);
2053					mutex_enter(&fd->fd_lock);
2054					if (error)
2055						break;
2056					dp += elem_size;
2057					len -= elem_size;
2058				} else {
2059					mutex_exit(&fp->f_lock);
2060					mutex_exit(&ff->ff_lock);
2061				}
2062				if (elem_count > 0) {
2063					needed += elem_size;
2064					if (elem_count != INT_MAX)
2065						elem_count--;
2066				}
2067			}
2068			mutex_exit(&fd->fd_lock);
2069
2070			/*
2071			 * Release reference to process.
2072			 */
2073			mutex_enter(proc_lock);
2074			rw_exit(&p->p_reflock);
2075		}
2076		mutex_exit(proc_lock);
2077		sysctl_relock();
2078		break;
2079	default:
2080		return (EINVAL);
2081	}
2082
2083	if (oldp == NULL)
2084		needed += KERN_FILESLOP * elem_size;
2085	*oldlenp = needed;
2086
2087	return (error);
2088}
2089
2090static void
2091fill_file(struct kinfo_file *kp, const file_t *fp, const fdfile_t *ff,
2092	  int i, pid_t pid)
2093{
2094
2095	memset(kp, 0, sizeof(*kp));
2096
2097	kp->ki_fileaddr =	PTRTOUINT64(fp);
2098	kp->ki_flag =		fp->f_flag;
2099	kp->ki_iflags =		fp->f_iflags;
2100	kp->ki_ftype =		fp->f_type;
2101	kp->ki_count =		fp->f_count;
2102	kp->ki_msgcount =	fp->f_msgcount;
2103	kp->ki_fucred =		PTRTOUINT64(fp->f_cred);
2104	kp->ki_fuid =		kauth_cred_geteuid(fp->f_cred);
2105	kp->ki_fgid =		kauth_cred_getegid(fp->f_cred);
2106	kp->ki_fops =		PTRTOUINT64(fp->f_ops);
2107	kp->ki_foffset =	fp->f_offset;
2108	kp->ki_fdata =		PTRTOUINT64(fp->f_data);
2109
2110	/* vnode information to glue this file to something */
2111	if (fp->f_type == DTYPE_VNODE) {
2112		struct vnode *vp = (struct vnode *)fp->f_data;
2113
2114		kp->ki_vun =	PTRTOUINT64(vp->v_un.vu_socket);
2115		kp->ki_vsize =	vp->v_size;
2116		kp->ki_vtype =	vp->v_type;
2117		kp->ki_vtag =	vp->v_tag;
2118		kp->ki_vdata =	PTRTOUINT64(vp->v_data);
2119	}
2120
2121	/* process information when retrieved via KERN_FILE_BYPID */
2122	if (ff != NULL) {
2123		kp->ki_pid =		pid;
2124		kp->ki_fd =		i;
2125		kp->ki_ofileflags =	ff->ff_exclose;
2126		kp->ki_usecount =	ff->ff_refcnt;
2127	}
2128}
2129
2130static int
2131sysctl_doeproc(SYSCTLFN_ARGS)
2132{
2133	struct eproc *eproc;
2134	struct kinfo_proc2 *kproc2;
2135	struct kinfo_proc *dp;
2136	struct proc *p, *next, *marker;
2137	char *where, *dp2;
2138	int type, op, arg, error;
2139	u_int elem_size, elem_count;
2140	size_t buflen, needed;
2141	bool match, zombie, mmmbrains;
2142
2143	if (namelen == 1 && name[0] == CTL_QUERY)
2144		return (sysctl_query(SYSCTLFN_CALL(rnode)));
2145
2146	dp = oldp;
2147	dp2 = where = oldp;
2148	buflen = where != NULL ? *oldlenp : 0;
2149	error = 0;
2150	needed = 0;
2151	type = rnode->sysctl_num;
2152
2153	if (type == KERN_PROC) {
2154		if (namelen != 2 && !(namelen == 1 && name[0] == KERN_PROC_ALL))
2155			return (EINVAL);
2156		op = name[0];
2157		if (op != KERN_PROC_ALL)
2158			arg = name[1];
2159		else
2160			arg = 0;		/* Quell compiler warning */
2161		elem_size = elem_count = 0;	/* Ditto */
2162	} else {
2163		if (namelen != 4)
2164			return (EINVAL);
2165		op = name[0];
2166		arg = name[1];
2167		elem_size = name[2];
2168		elem_count = name[3];
2169	}
2170
2171	sysctl_unlock();
2172
2173	if (type == KERN_PROC) {
2174		eproc = kmem_alloc(sizeof(*eproc), KM_SLEEP);
2175		kproc2 = NULL;
2176	} else {
2177		eproc = NULL;
2178		kproc2 = kmem_alloc(sizeof(*kproc2), KM_SLEEP);
2179	}
2180	marker = kmem_alloc(sizeof(*marker), KM_SLEEP);
2181
2182	mutex_enter(proc_lock);
2183	mmmbrains = false;
2184	for (p = LIST_FIRST(&allproc);; p = next) {
2185		if (p == NULL) {
2186			if (!mmmbrains) {
2187				p = LIST_FIRST(&zombproc);
2188				mmmbrains = true;
2189			}
2190			if (p == NULL)
2191				break;
2192		}
2193		next = LIST_NEXT(p, p_list);
2194
2195		/*
2196		 * Skip embryonic processes.
2197		 */
2198		if (p->p_stat == SIDL)
2199			continue;
2200
2201		mutex_enter(p->p_lock);
2202		error = kauth_authorize_process(l->l_cred,
2203		    KAUTH_PROCESS_CANSEE, p,
2204		    KAUTH_ARG(KAUTH_REQ_PROCESS_CANSEE_ENTRY), NULL, NULL);
2205		if (error != 0) {
2206			mutex_exit(p->p_lock);
2207			continue;
2208		}
2209
2210		/*
2211		 * TODO - make more efficient (see notes below).
2212		 * do by session.
2213		 */
2214		switch (op) {
2215		case KERN_PROC_PID:
2216			/* could do this with just a lookup */
2217			match = (p->p_pid == (pid_t)arg);
2218			break;
2219
2220		case KERN_PROC_PGRP:
2221			/* could do this by traversing pgrp */
2222			match = (p->p_pgrp->pg_id == (pid_t)arg);
2223			break;
2224
2225		case KERN_PROC_SESSION:
2226			match = (p->p_session->s_sid == (pid_t)arg);
2227			break;
2228
2229		case KERN_PROC_TTY:
2230			match = true;
2231			if (arg == (int) KERN_PROC_TTY_REVOKE) {
2232				if ((p->p_lflag & PL_CONTROLT) == 0 ||
2233				    p->p_session->s_ttyp == NULL ||
2234				    p->p_session->s_ttyvp != NULL) {
2235				    	match = false;
2236				}
2237			} else if ((p->p_lflag & PL_CONTROLT) == 0 ||
2238			    p->p_session->s_ttyp == NULL) {
2239				if ((dev_t)arg != KERN_PROC_TTY_NODEV) {
2240					match = false;
2241				}
2242			} else if (p->p_session->s_ttyp->t_dev != (dev_t)arg) {
2243				match = false;
2244			}
2245			break;
2246
2247		case KERN_PROC_UID:
2248			match = (kauth_cred_geteuid(p->p_cred) == (uid_t)arg);
2249			break;
2250
2251		case KERN_PROC_RUID:
2252			match = (kauth_cred_getuid(p->p_cred) == (uid_t)arg);
2253			break;
2254
2255		case KERN_PROC_GID:
2256			match = (kauth_cred_getegid(p->p_cred) == (uid_t)arg);
2257			break;
2258
2259		case KERN_PROC_RGID:
2260			match = (kauth_cred_getgid(p->p_cred) == (uid_t)arg);
2261			break;
2262
2263		case KERN_PROC_ALL:
2264			match = true;
2265			/* allow everything */
2266			break;
2267
2268		default:
2269			error = EINVAL;
2270			mutex_exit(p->p_lock);
2271			goto cleanup;
2272		}
2273		if (!match) {
2274			mutex_exit(p->p_lock);
2275			continue;
2276		}
2277
2278		/*
2279		 * Grab a hold on the process.
2280		 */
2281		if (mmmbrains) {
2282			zombie = true;
2283		} else {
2284			zombie = !rw_tryenter(&p->p_reflock, RW_READER);
2285		}
2286		if (zombie) {
2287			LIST_INSERT_AFTER(p, marker, p_list);
2288		}
2289
2290		if (type == KERN_PROC) {
2291			if (buflen >= sizeof(struct kinfo_proc)) {
2292				fill_eproc(p, eproc, zombie);
2293				mutex_exit(p->p_lock);
2294				mutex_exit(proc_lock);
2295				error = dcopyout(l, p, &dp->kp_proc,
2296				    sizeof(struct proc));
2297				mutex_enter(proc_lock);
2298				if (error) {
2299					goto bah;
2300				}
2301				error = dcopyout(l, eproc, &dp->kp_eproc,
2302				    sizeof(*eproc));
2303				if (error) {
2304					goto bah;
2305				}
2306				dp++;
2307				buflen -= sizeof(struct kinfo_proc);
2308			} else {
2309				mutex_exit(p->p_lock);
2310			}
2311			needed += sizeof(struct kinfo_proc);
2312		} else { /* KERN_PROC2 */
2313			if (buflen >= elem_size && elem_count > 0) {
2314				fill_kproc2(p, kproc2, zombie);
2315				mutex_exit(p->p_lock);
2316				mutex_exit(proc_lock);
2317				/*
2318				 * Copy out elem_size, but not larger than
2319				 * the size of a struct kinfo_proc2.
2320				 */
2321				error = dcopyout(l, kproc2, dp2,
2322				    min(sizeof(*kproc2), elem_size));
2323				mutex_enter(proc_lock);
2324				if (error) {
2325					goto bah;
2326				}
2327				dp2 += elem_size;
2328				buflen -= elem_size;
2329				elem_count--;
2330			} else {
2331				mutex_exit(p->p_lock);
2332			}
2333			needed += elem_size;
2334		}
2335
2336		/*
2337		 * Release reference to process.
2338		 */
2339	 	if (zombie) {
2340			next = LIST_NEXT(marker, p_list);
2341 			LIST_REMOVE(marker, p_list);
2342		} else {
2343			rw_exit(&p->p_reflock);
2344		}
2345	}
2346	mutex_exit(proc_lock);
2347
2348	if (where != NULL) {
2349		if (type == KERN_PROC)
2350			*oldlenp = (char *)dp - where;
2351		else
2352			*oldlenp = dp2 - where;
2353		if (needed > *oldlenp) {
2354			error = ENOMEM;
2355			goto out;
2356		}
2357	} else {
2358		needed += KERN_PROCSLOP;
2359		*oldlenp = needed;
2360	}
2361	if (kproc2)
2362		kmem_free(kproc2, sizeof(*kproc2));
2363	if (eproc)
2364		kmem_free(eproc, sizeof(*eproc));
2365	if (marker)
2366		kmem_free(marker, sizeof(*marker));
2367	sysctl_relock();
2368	return 0;
2369 bah:
2370 	if (zombie)
2371 		LIST_REMOVE(marker, p_list);
2372	else
2373		rw_exit(&p->p_reflock);
2374 cleanup:
2375	mutex_exit(proc_lock);
2376 out:
2377	if (kproc2)
2378		kmem_free(kproc2, sizeof(*kproc2));
2379	if (eproc)
2380		kmem_free(eproc, sizeof(*eproc));
2381	if (marker)
2382		kmem_free(marker, sizeof(*marker));
2383	sysctl_relock();
2384	return error;
2385}
2386
2387/*
2388 * sysctl helper routine for kern.proc_args pseudo-subtree.
2389 */
2390static int
2391sysctl_kern_proc_args(SYSCTLFN_ARGS)
2392{
2393	struct ps_strings pss;
2394	struct proc *p;
2395	size_t len, i;
2396	struct uio auio;
2397	struct iovec aiov;
2398	pid_t pid;
2399	int nargv, type, error, argvlen;
2400	char *arg;
2401	char **argv = NULL;
2402	char *tmp;
2403	struct vmspace *vmspace;
2404	vaddr_t psstr_addr;
2405	vaddr_t offsetn;
2406	vaddr_t offsetv;
2407
2408	if (namelen == 1 && name[0] == CTL_QUERY)
2409		return (sysctl_query(SYSCTLFN_CALL(rnode)));
2410
2411	if (newp != NULL || namelen != 2)
2412		return (EINVAL);
2413	pid = name[0];
2414	type = name[1];
2415	argv = NULL;
2416	argvlen = 0;
2417
2418	switch (type) {
2419	case KERN_PROC_ARGV:
2420	case KERN_PROC_NARGV:
2421	case KERN_PROC_ENV:
2422	case KERN_PROC_NENV:
2423		/* ok */
2424		break;
2425	default:
2426		return (EINVAL);
2427	}
2428
2429	sysctl_unlock();
2430
2431	/* check pid */
2432	mutex_enter(proc_lock);
2433	if ((p = p_find(pid, PFIND_LOCKED)) == NULL) {
2434		error = EINVAL;
2435		goto out_locked;
2436	}
2437	mutex_enter(p->p_lock);
2438
2439	/* Check permission. */
2440	if (type == KERN_PROC_ARGV || type == KERN_PROC_NARGV)
2441		error = kauth_authorize_process(l->l_cred, KAUTH_PROCESS_CANSEE,
2442		    p, KAUTH_ARG(KAUTH_REQ_PROCESS_CANSEE_ARGS), NULL, NULL);
2443	else if (type == KERN_PROC_ENV || type == KERN_PROC_NENV)
2444		error = kauth_authorize_process(l->l_cred, KAUTH_PROCESS_CANSEE,
2445		    p, KAUTH_ARG(KAUTH_REQ_PROCESS_CANSEE_ENV), NULL, NULL);
2446	else
2447		error = EINVAL; /* XXXGCC */
2448	if (error) {
2449		mutex_exit(p->p_lock);
2450		goto out_locked;
2451	}
2452
2453	if (oldp == NULL) {
2454		if (type == KERN_PROC_NARGV || type == KERN_PROC_NENV)
2455			*oldlenp = sizeof (int);
2456		else
2457			*oldlenp = ARG_MAX;	/* XXX XXX XXX */
2458		error = 0;
2459		mutex_exit(p->p_lock);
2460		goto out_locked;
2461	}
2462
2463	/*
2464	 * Zombies don't have a stack, so we can't read their psstrings.
2465	 * System processes also don't have a user stack.
2466	 */
2467	if (P_ZOMBIE(p) || (p->p_flag & PK_SYSTEM) != 0) {
2468		error = EINVAL;
2469		mutex_exit(p->p_lock);
2470		goto out_locked;
2471	}
2472
2473	/*
2474	 * Lock the process down in memory.
2475	 */
2476	psstr_addr = (vaddr_t)p->p_psstr;
2477	if (type == KERN_PROC_ARGV || type == KERN_PROC_NARGV) {
2478		offsetn = p->p_psnargv;
2479		offsetv = p->p_psargv;
2480	} else {
2481		offsetn = p->p_psnenv;
2482		offsetv = p->p_psenv;
2483	}
2484	vmspace = p->p_vmspace;
2485	uvmspace_addref(vmspace);
2486	mutex_exit(p->p_lock);
2487	mutex_exit(proc_lock);
2488
2489	/*
2490	 * Allocate a temporary buffer to hold the arguments.
2491	 */
2492	arg = kmem_alloc(PAGE_SIZE, KM_SLEEP);
2493
2494	/*
2495	 * Read in the ps_strings structure.
2496	 */
2497	aiov.iov_base = &pss;
2498	aiov.iov_len = sizeof(pss);
2499	auio.uio_iov = &aiov;
2500	auio.uio_iovcnt = 1;
2501	auio.uio_offset = psstr_addr;
2502	auio.uio_resid = sizeof(pss);
2503	auio.uio_rw = UIO_READ;
2504	UIO_SETUP_SYSSPACE(&auio);
2505	error = uvm_io(&vmspace->vm_map, &auio);
2506	if (error)
2507		goto done;
2508
2509	memcpy(&nargv, (char *)&pss + offsetn, sizeof(nargv));
2510	if (type == KERN_PROC_NARGV || type == KERN_PROC_NENV) {
2511		error = dcopyout(l, &nargv, oldp, sizeof(nargv));
2512		*oldlenp = sizeof(nargv);
2513		goto done;
2514	}
2515	/*
2516	 * Now read the address of the argument vector.
2517	 */
2518	switch (type) {
2519	case KERN_PROC_ARGV:
2520		/* FALLTHROUGH */
2521	case KERN_PROC_ENV:
2522		memcpy(&tmp, (char *)&pss + offsetv, sizeof(tmp));
2523		break;
2524	default:
2525		error = EINVAL;
2526		goto done;
2527	}
2528
2529#ifdef COMPAT_NETBSD32
2530	if (p->p_flag & PK_32)
2531		len = sizeof(netbsd32_charp) * nargv;
2532	else
2533#endif
2534		len = sizeof(char *) * nargv;
2535
2536	if ((argvlen = len) != 0)
2537		argv = kmem_alloc(len, KM_SLEEP);
2538
2539	aiov.iov_base = argv;
2540	aiov.iov_len = len;
2541	auio.uio_iov = &aiov;
2542	auio.uio_iovcnt = 1;
2543	auio.uio_offset = (off_t)(unsigned long)tmp;
2544	auio.uio_resid = len;
2545	auio.uio_rw = UIO_READ;
2546	UIO_SETUP_SYSSPACE(&auio);
2547	error = uvm_io(&vmspace->vm_map, &auio);
2548	if (error)
2549		goto done;
2550
2551	/*
2552	 * Now copy each string.
2553	 */
2554	len = 0; /* bytes written to user buffer */
2555	for (i = 0; i < nargv; i++) {
2556		int finished = 0;
2557		vaddr_t base;
2558		size_t xlen;
2559		int j;
2560
2561#ifdef COMPAT_NETBSD32
2562		if (p->p_flag & PK_32) {
2563			netbsd32_charp *argv32;
2564
2565			argv32 = (netbsd32_charp *)argv;
2566			base = (vaddr_t)NETBSD32PTR64(argv32[i]);
2567		} else
2568#endif
2569			base = (vaddr_t)argv[i];
2570
2571		/*
2572		 * The program has messed around with its arguments,
2573		 * possibly deleting some, and replacing them with
2574		 * NULL's. Treat this as the last argument and not
2575		 * a failure.
2576		 */
2577		if (base == 0)
2578			break;
2579
2580		while (!finished) {
2581			xlen = PAGE_SIZE - (base & PAGE_MASK);
2582
2583			aiov.iov_base = arg;
2584			aiov.iov_len = PAGE_SIZE;
2585			auio.uio_iov = &aiov;
2586			auio.uio_iovcnt = 1;
2587			auio.uio_offset = base;
2588			auio.uio_resid = xlen;
2589			auio.uio_rw = UIO_READ;
2590			UIO_SETUP_SYSSPACE(&auio);
2591			error = uvm_io(&vmspace->vm_map, &auio);
2592			if (error)
2593				goto done;
2594
2595			/* Look for the end of the string */
2596			for (j = 0; j < xlen; j++) {
2597				if (arg[j] == '\0') {
2598					xlen = j + 1;
2599					finished = 1;
2600					break;
2601				}
2602			}
2603
2604			/* Check for user buffer overflow */
2605			if (len + xlen > *oldlenp) {
2606				finished = 1;
2607				if (len > *oldlenp)
2608					xlen = 0;
2609				else
2610					xlen = *oldlenp - len;
2611			}
2612
2613			/* Copyout the page */
2614			error = dcopyout(l, arg, (char *)oldp + len, xlen);
2615			if (error)
2616				goto done;
2617
2618			len += xlen;
2619			base += xlen;
2620		}
2621	}
2622	*oldlenp = len;
2623
2624done:
2625	if (argvlen != 0)
2626		kmem_free(argv, argvlen);
2627	uvmspace_free(vmspace);
2628	kmem_free(arg, PAGE_SIZE);
2629	sysctl_relock();
2630	return error;
2631
2632out_locked:
2633	mutex_exit(proc_lock);
2634	sysctl_relock();
2635	return error;
2636}
2637
2638static int
2639sysctl_security_setidcore(SYSCTLFN_ARGS)
2640{
2641	int newsize, error;
2642	struct sysctlnode node;
2643
2644	node = *rnode;
2645	node.sysctl_data = &newsize;
2646	newsize = *(int *)rnode->sysctl_data;
2647	error = sysctl_lookup(SYSCTLFN_CALL(&node));
2648	if (error || newp == NULL)
2649		return error;
2650
2651	if (kauth_authorize_system(l->l_cred, KAUTH_SYSTEM_SETIDCORE,
2652	    0, NULL, NULL, NULL))
2653		return (EPERM);
2654
2655	*(int *)rnode->sysctl_data = newsize;
2656
2657	return 0;
2658}
2659
2660static int
2661sysctl_security_setidcorename(SYSCTLFN_ARGS)
2662{
2663	int error;
2664	char *newsetidcorename;
2665	struct sysctlnode node;
2666
2667	newsetidcorename = PNBUF_GET();
2668	node = *rnode;
2669	node.sysctl_data = newsetidcorename;
2670	memcpy(node.sysctl_data, rnode->sysctl_data, MAXPATHLEN);
2671	error = sysctl_lookup(SYSCTLFN_CALL(&node));
2672	if (error || newp == NULL) {
2673		goto out;
2674	}
2675	if (kauth_authorize_system(l->l_cred, KAUTH_SYSTEM_SETIDCORE,
2676	    0, NULL, NULL, NULL)) {
2677		error = EPERM;
2678		goto out;
2679	}
2680	if (strlen(newsetidcorename) == 0) {
2681		error = EINVAL;
2682		goto out;
2683	}
2684	memcpy(rnode->sysctl_data, node.sysctl_data, MAXPATHLEN);
2685out:
2686	PNBUF_PUT(newsetidcorename);
2687	return error;
2688}
2689
2690/*
2691 * sysctl helper routine for kern.cp_id node. Maps cpus to their
2692 * cpuids.
2693 */
2694static int
2695sysctl_kern_cpid(SYSCTLFN_ARGS)
2696{
2697	struct sysctlnode node = *rnode;
2698	uint64_t *cp_id = NULL;
2699	int error, n = ncpu;
2700	struct cpu_info *ci;
2701	CPU_INFO_ITERATOR cii;
2702
2703	/*
2704	 * Here you may either retrieve a single cpu id or the whole
2705	 * set. The size you get back when probing depends on what
2706	 * you ask for.
2707	 */
2708	switch (namelen) {
2709	case 0:
2710		node.sysctl_size = n * sizeof(uint64_t);
2711		n = -2; /* ALL */
2712		break;
2713	case 1:
2714		if (name[0] < 0 || name[0] >= n)
2715			return (ENOENT); /* ENOSUCHPROCESSOR */
2716		node.sysctl_size = sizeof(uint64_t);
2717		n = name[0];
2718		/*
2719		 * adjust these so that sysctl_lookup() will be happy
2720		 */
2721		name++;
2722		namelen--;
2723		break;
2724	default:
2725		return (EINVAL);
2726	}
2727
2728	cp_id = kmem_alloc(node.sysctl_size, KM_SLEEP);
2729	if (cp_id == NULL)
2730		return (ENOMEM);
2731	node.sysctl_data = cp_id;
2732	memset(cp_id, 0, node.sysctl_size);
2733
2734	for (CPU_INFO_FOREACH(cii, ci)) {
2735		if (n <= 0)
2736			cp_id[0] = ci->ci_cpuid;
2737		/*
2738		 * if a specific processor was requested and we just
2739		 * did it, we're done here
2740		 */
2741		if (n == 0)
2742			break;
2743		/*
2744		 * if doing "all", skip to next cp_id slot for next processor
2745		 */
2746		if (n == -2)
2747			cp_id++;
2748		/*
2749		 * if we're doing a specific processor, we're one
2750		 * processor closer
2751		 */
2752		if (n > 0)
2753			n--;
2754	}
2755
2756	error = sysctl_lookup(SYSCTLFN_CALL(&node));
2757	kmem_free(node.sysctl_data, node.sysctl_size);
2758	return (error);
2759}
2760
2761/*
2762 * sysctl helper routine for hw.usermem and hw.usermem64. Values are
2763 * calculate on the fly taking into account integer overflow and the
2764 * current wired count.
2765 */
2766static int
2767sysctl_hw_usermem(SYSCTLFN_ARGS)
2768{
2769	u_int ui;
2770	u_quad_t uq;
2771	struct sysctlnode node;
2772
2773	node = *rnode;
2774	switch (rnode->sysctl_num) {
2775	    case HW_USERMEM:
2776		if ((ui = physmem - uvmexp.wired) > (UINT_MAX / PAGE_SIZE))
2777			ui = UINT_MAX;
2778		else
2779			ui *= PAGE_SIZE;
2780		node.sysctl_data = &ui;
2781		break;
2782	case HW_USERMEM64:
2783		uq = (u_quad_t)(physmem - uvmexp.wired) * PAGE_SIZE;
2784		node.sysctl_data = &uq;
2785		break;
2786	default:
2787		return (EINVAL);
2788	}
2789
2790	return (sysctl_lookup(SYSCTLFN_CALL(&node)));
2791}
2792
2793/*
2794 * sysctl helper routine for kern.cnmagic node. Pulls the old value
2795 * out, encoded, and stuffs the new value in for decoding.
2796 */
2797static int
2798sysctl_hw_cnmagic(SYSCTLFN_ARGS)
2799{
2800	char magic[CNS_LEN];
2801	int error;
2802	struct sysctlnode node;
2803
2804	if (oldp)
2805		cn_get_magic(magic, CNS_LEN);
2806	node = *rnode;
2807	node.sysctl_data = &magic[0];
2808	error = sysctl_lookup(SYSCTLFN_CALL(&node));
2809	if (error || newp == NULL)
2810		return (error);
2811
2812	return (cn_set_magic(magic));
2813}
2814
2815/*
2816 * ********************************************************************
2817 * section 3: public helper routines that are used for more than one
2818 * node
2819 * ********************************************************************
2820 */
2821
2822/*
2823 * sysctl helper routine for the kern.root_device node and some ports'
2824 * machdep.root_device nodes.
2825 */
2826int
2827sysctl_root_device(SYSCTLFN_ARGS)
2828{
2829	struct sysctlnode node;
2830
2831	node = *rnode;
2832	node.sysctl_data = root_device->dv_xname;
2833	node.sysctl_size = strlen(device_xname(root_device)) + 1;
2834	return (sysctl_lookup(SYSCTLFN_CALL(&node)));
2835}
2836
2837/*
2838 * sysctl helper routine for kern.consdev, dependent on the current
2839 * state of the console. Also used for machdep.console_device on some
2840 * ports.
2841 */
2842int
2843sysctl_consdev(SYSCTLFN_ARGS)
2844{
2845	dev_t consdev;
2846	struct sysctlnode node;
2847
2848	if (cn_tab != NULL)
2849		consdev = cn_tab->cn_dev;
2850	else
2851		consdev = NODEV;
2852	node = *rnode;
2853	node.sysctl_data = &consdev;
2854	node.sysctl_size = sizeof(consdev);
2855	return (sysctl_lookup(SYSCTLFN_CALL(&node)));
2856}
2857
2858/*
2859 * ********************************************************************
2860 * section 4: support for some helpers
2861 * ********************************************************************
2862 */
2863
2864/*
2865 * Fill in a kinfo_proc2 structure for the specified process.
2866 */
2867static void
2868fill_kproc2(struct proc *p, struct kinfo_proc2 *ki, bool zombie)
2869{
2870	struct tty *tp;
2871	struct lwp *l, *l2;
2872	struct timeval ut, st, rt;
2873	sigset_t ss1, ss2;
2874	struct rusage ru;
2875	struct vmspace *vm;
2876	int tmp;
2877
2878	KASSERT(mutex_owned(proc_lock));
2879	KASSERT(mutex_owned(p->p_lock));
2880
2881	sigemptyset(&ss1);
2882	sigemptyset(&ss2);
2883	memset(ki, 0, sizeof(*ki));
2884
2885	ki->p_paddr = PTRTOUINT64(p);
2886	ki->p_fd = PTRTOUINT64(p->p_fd);
2887	ki->p_cwdi = PTRTOUINT64(p->p_cwdi);
2888	ki->p_stats = PTRTOUINT64(p->p_stats);
2889	ki->p_limit = PTRTOUINT64(p->p_limit);
2890	ki->p_vmspace = PTRTOUINT64(p->p_vmspace);
2891	ki->p_sigacts = PTRTOUINT64(p->p_sigacts);
2892	ki->p_sess = PTRTOUINT64(p->p_session);
2893	ki->p_tsess = 0;	/* may be changed if controlling tty below */
2894	ki->p_ru = PTRTOUINT64(&p->p_stats->p_ru);
2895	ki->p_eflag = 0;
2896	ki->p_exitsig = p->p_exitsig;
2897	ki->p_flag = sysctl_map_flags(sysctl_flagmap, p->p_flag);
2898	ki->p_flag |= sysctl_map_flags(sysctl_sflagmap, p->p_sflag);
2899	ki->p_flag |= sysctl_map_flags(sysctl_slflagmap, p->p_slflag);
2900	ki->p_flag |= sysctl_map_flags(sysctl_lflagmap, p->p_lflag);
2901	ki->p_flag |= sysctl_map_flags(sysctl_stflagmap, p->p_stflag);
2902	ki->p_pid = p->p_pid;
2903	if (p->p_pptr)
2904		ki->p_ppid = p->p_pptr->p_pid;
2905	else
2906		ki->p_ppid = 0;
2907	ki->p_uid = kauth_cred_geteuid(p->p_cred);
2908	ki->p_ruid = kauth_cred_getuid(p->p_cred);
2909	ki->p_gid = kauth_cred_getegid(p->p_cred);
2910	ki->p_rgid = kauth_cred_getgid(p->p_cred);
2911	ki->p_svuid = kauth_cred_getsvuid(p->p_cred);
2912	ki->p_svgid = kauth_cred_getsvgid(p->p_cred);
2913	ki->p_ngroups = kauth_cred_ngroups(p->p_cred);
2914	kauth_cred_getgroups(p->p_cred, ki->p_groups,
2915	    min(ki->p_ngroups, sizeof(ki->p_groups) / sizeof(ki->p_groups[0])),
2916	    UIO_SYSSPACE);
2917
2918	ki->p_uticks = p->p_uticks;
2919	ki->p_sticks = p->p_sticks;
2920	ki->p_iticks = p->p_iticks;
2921	ki->p_tpgid = NO_PGID;	/* may be changed if controlling tty below */
2922	ki->p_tracep = PTRTOUINT64(p->p_tracep);
2923	ki->p_traceflag = p->p_traceflag;
2924
2925	memcpy(&ki->p_sigignore, &p->p_sigctx.ps_sigignore,sizeof(ki_sigset_t));
2926	memcpy(&ki->p_sigcatch, &p->p_sigctx.ps_sigcatch, sizeof(ki_sigset_t));
2927
2928	ki->p_cpticks = 0;
2929	ki->p_pctcpu = p->p_pctcpu;
2930	ki->p_estcpu = 0;
2931	ki->p_stat = p->p_stat; /* Will likely be overridden by LWP status */
2932	ki->p_realstat = p->p_stat;
2933	ki->p_nice = p->p_nice;
2934	ki->p_xstat = p->p_xstat;
2935	ki->p_acflag = p->p_acflag;
2936
2937	strncpy(ki->p_comm, p->p_comm,
2938	    min(sizeof(ki->p_comm), sizeof(p->p_comm)));
2939
2940	ki->p_nlwps = p->p_nlwps;
2941	ki->p_realflag = ki->p_flag;
2942
2943	if (p->p_stat != SIDL && !P_ZOMBIE(p) && !zombie) {
2944		vm = p->p_vmspace;
2945		ki->p_vm_rssize = vm_resident_count(vm);
2946		ki->p_vm_tsize = vm->vm_tsize;
2947		ki->p_vm_dsize = vm->vm_dsize;
2948		ki->p_vm_ssize = vm->vm_ssize;
2949
2950		/* Pick a "representative" LWP */
2951		l = proc_representative_lwp(p, &tmp, 1);
2952		lwp_lock(l);
2953		ki->p_nrlwps = tmp;
2954		ki->p_forw = 0;
2955		ki->p_back = 0;
2956		ki->p_addr = PTRTOUINT64(l->l_addr);
2957		ki->p_stat = l->l_stat;
2958		ki->p_flag |= sysctl_map_flags(sysctl_lwpflagmap, l->l_flag);
2959		ki->p_swtime = l->l_swtime;
2960		ki->p_slptime = l->l_slptime;
2961		if (l->l_stat == LSONPROC)
2962			ki->p_schedflags = l->l_cpu->ci_schedstate.spc_flags;
2963		else
2964			ki->p_schedflags = 0;
2965		ki->p_holdcnt = l->l_holdcnt;
2966		ki->p_priority = lwp_eprio(l);
2967		ki->p_usrpri = l->l_priority;
2968		if (l->l_wchan)
2969			strncpy(ki->p_wmesg, l->l_wmesg, sizeof(ki->p_wmesg));
2970		ki->p_wchan = PTRTOUINT64(l->l_wchan);
2971		ki->p_cpuid = l->l_cpu->ci_cpuid;
2972		lwp_unlock(l);
2973		LIST_FOREACH(l, &p->p_lwps, l_sibling) {
2974			/* This is hardly correct, but... */
2975			sigplusset(&l->l_sigpend.sp_set, &ss1);
2976			sigplusset(&l->l_sigmask, &ss2);
2977			ki->p_cpticks += l->l_cpticks;
2978			ki->p_pctcpu += l->l_pctcpu;
2979			ki->p_estcpu += l->l_estcpu;
2980		}
2981	}
2982	sigplusset(&p->p_sigpend.sp_set, &ss2);
2983	memcpy(&ki->p_siglist, &ss1, sizeof(ki_sigset_t));
2984	memcpy(&ki->p_sigmask, &ss2, sizeof(ki_sigset_t));
2985
2986	if (p->p_session != NULL) {
2987		ki->p_sid = p->p_session->s_sid;
2988		ki->p__pgid = p->p_pgrp->pg_id;
2989		if (p->p_session->s_ttyvp)
2990			ki->p_eflag |= EPROC_CTTY;
2991		if (SESS_LEADER(p))
2992			ki->p_eflag |= EPROC_SLEADER;
2993		strncpy(ki->p_login, p->p_session->s_login,
2994		    min(sizeof ki->p_login - 1, sizeof p->p_session->s_login));
2995		ki->p_jobc = p->p_pgrp->pg_jobc;
2996		if ((p->p_lflag & PL_CONTROLT) && (tp = p->p_session->s_ttyp)) {
2997			ki->p_tdev = tp->t_dev;
2998			ki->p_tpgid = tp->t_pgrp ? tp->t_pgrp->pg_id : NO_PGID;
2999			ki->p_tsess = PTRTOUINT64(tp->t_session);
3000		} else {
3001			ki->p_tdev = NODEV;
3002		}
3003	}
3004
3005	if (!P_ZOMBIE(p) && !zombie) {
3006		ki->p_uvalid = 1;
3007		ki->p_ustart_sec = p->p_stats->p_start.tv_sec;
3008		ki->p_ustart_usec = p->p_stats->p_start.tv_usec;
3009
3010		calcru(p, &ut, &st, NULL, &rt);
3011		ki->p_rtime_sec = rt.tv_sec;
3012		ki->p_rtime_usec = rt.tv_usec;
3013		ki->p_uutime_sec = ut.tv_sec;
3014		ki->p_uutime_usec = ut.tv_usec;
3015		ki->p_ustime_sec = st.tv_sec;
3016		ki->p_ustime_usec = st.tv_usec;
3017
3018		memcpy(&ru, &p->p_stats->p_ru, sizeof(ru));
3019		ki->p_uru_nvcsw = 0;
3020		ki->p_uru_nivcsw = 0;
3021		LIST_FOREACH(l2, &p->p_lwps, l_sibling) {
3022			ki->p_uru_nvcsw += (l2->l_ncsw - l2->l_nivcsw);
3023			ki->p_uru_nivcsw += l2->l_nivcsw;
3024			ruadd(&ru, &l2->l_ru);
3025		}
3026		ki->p_uru_maxrss = ru.ru_maxrss;
3027		ki->p_uru_ixrss = ru.ru_ixrss;
3028		ki->p_uru_idrss = ru.ru_idrss;
3029		ki->p_uru_isrss = ru.ru_isrss;
3030		ki->p_uru_minflt = ru.ru_minflt;
3031		ki->p_uru_majflt = ru.ru_majflt;
3032		ki->p_uru_nswap = ru.ru_nswap;
3033		ki->p_uru_inblock = ru.ru_inblock;
3034		ki->p_uru_oublock = ru.ru_oublock;
3035		ki->p_uru_msgsnd = ru.ru_msgsnd;
3036		ki->p_uru_msgrcv = ru.ru_msgrcv;
3037		ki->p_uru_nsignals = ru.ru_nsignals;
3038
3039		timeradd(&p->p_stats->p_cru.ru_utime,
3040			 &p->p_stats->p_cru.ru_stime, &ut);
3041		ki->p_uctime_sec = ut.tv_sec;
3042		ki->p_uctime_usec = ut.tv_usec;
3043	}
3044}
3045
3046/*
3047 * Fill in a kinfo_lwp structure for the specified lwp.
3048 */
3049static void
3050fill_lwp(struct lwp *l, struct kinfo_lwp *kl)
3051{
3052	struct proc *p = l->l_proc;
3053	struct timeval tv;
3054
3055	KASSERT(lwp_locked(l, NULL));
3056
3057	kl->l_forw = 0;
3058	kl->l_back = 0;
3059	kl->l_laddr = PTRTOUINT64(l);
3060	kl->l_addr = PTRTOUINT64(l->l_addr);
3061	kl->l_stat = l->l_stat;
3062	kl->l_lid = l->l_lid;
3063	kl->l_flag = sysctl_map_flags(sysctl_lwpprflagmap, l->l_prflag);
3064
3065	kl->l_swtime = l->l_swtime;
3066	kl->l_slptime = l->l_slptime;
3067	if (l->l_stat == LSONPROC)
3068		kl->l_schedflags = l->l_cpu->ci_schedstate.spc_flags;
3069	else
3070		kl->l_schedflags = 0;
3071	kl->l_holdcnt = l->l_holdcnt;
3072	kl->l_priority = lwp_eprio(l);
3073	kl->l_usrpri = l->l_priority;
3074	if (l->l_wchan)
3075		strncpy(kl->l_wmesg, l->l_wmesg, sizeof(kl->l_wmesg));
3076	kl->l_wchan = PTRTOUINT64(l->l_wchan);
3077	kl->l_cpuid = l->l_cpu->ci_cpuid;
3078	bintime2timeval(&l->l_rtime, &tv);
3079	kl->l_rtime_sec = tv.tv_sec;
3080	kl->l_rtime_usec = tv.tv_usec;
3081	kl->l_cpticks = l->l_cpticks;
3082	kl->l_pctcpu = l->l_pctcpu;
3083	kl->l_pid = p->p_pid;
3084	if (l->l_name == NULL)
3085		kl->l_name[0] = '\0';
3086	else
3087		strlcpy(kl->l_name, l->l_name, sizeof(kl->l_name));
3088}
3089
3090/*
3091 * Fill in an eproc structure for the specified process.
3092 */
3093void
3094fill_eproc(struct proc *p, struct eproc *ep, bool zombie)
3095{
3096	struct tty *tp;
3097	struct lwp *l;
3098
3099	KASSERT(mutex_owned(proc_lock));
3100	KASSERT(mutex_owned(p->p_lock));
3101
3102	memset(ep, 0, sizeof(*ep));
3103
3104	ep->e_paddr = p;
3105	ep->e_sess = p->p_session;
3106	if (p->p_cred) {
3107		kauth_cred_topcred(p->p_cred, &ep->e_pcred);
3108		kauth_cred_toucred(p->p_cred, &ep->e_ucred);
3109	}
3110	if (p->p_stat != SIDL && !P_ZOMBIE(p) && !zombie) {
3111		struct vmspace *vm = p->p_vmspace;
3112
3113		ep->e_vm.vm_rssize = vm_resident_count(vm);
3114		ep->e_vm.vm_tsize = vm->vm_tsize;
3115		ep->e_vm.vm_dsize = vm->vm_dsize;
3116		ep->e_vm.vm_ssize = vm->vm_ssize;
3117
3118		/* Pick a "representative" LWP */
3119		l = proc_representative_lwp(p, NULL, 1);
3120		lwp_lock(l);
3121		if (l->l_wchan)
3122			strncpy(ep->e_wmesg, l->l_wmesg, WMESGLEN);
3123		lwp_unlock(l);
3124	}
3125	if (p->p_pptr)
3126		ep->e_ppid = p->p_pptr->p_pid;
3127	if (p->p_pgrp && p->p_session) {
3128		ep->e_pgid = p->p_pgrp->pg_id;
3129		ep->e_jobc = p->p_pgrp->pg_jobc;
3130		ep->e_sid = p->p_session->s_sid;
3131		if ((p->p_lflag & PL_CONTROLT) &&
3132		    (tp = ep->e_sess->s_ttyp)) {
3133			ep->e_tdev = tp->t_dev;
3134			ep->e_tpgid = tp->t_pgrp ? tp->t_pgrp->pg_id : NO_PGID;
3135			ep->e_tsess = tp->t_session;
3136		} else
3137			ep->e_tdev = NODEV;
3138		ep->e_flag = ep->e_sess->s_ttyvp ? EPROC_CTTY : 0;
3139		if (SESS_LEADER(p))
3140			ep->e_flag |= EPROC_SLEADER;
3141		strncpy(ep->e_login, ep->e_sess->s_login, MAXLOGNAME);
3142	}
3143	ep->e_xsize = ep->e_xrssize = 0;
3144	ep->e_xccount = ep->e_xswrss = 0;
3145}
3146
3147u_int
3148sysctl_map_flags(const u_int *map, u_int word)
3149{
3150	u_int rv;
3151
3152	for (rv = 0; *map != 0; map += 2)
3153		if ((word & map[0]) != 0)
3154			rv |= map[1];
3155
3156	return rv;
3157}
3158