xref: /netbsd-src/sys/kern/init_sysctl.c (revision cac8e449158efc7261bebc8657cbb0125a2cfdde)
1 /*	$NetBSD: init_sysctl.c,v 1.144 2008/07/15 22:25:30 christos 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 1.144 2008/07/15 22:25:30 christos 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 #include <miscfs/specfs/specdev.h>
71 
72 #ifdef COMPAT_NETBSD32
73 #include <compat/netbsd32/netbsd32.h>
74 #endif
75 
76 #include <sys/cpu.h>
77 
78 /* XXX this should not be here */
79 int security_setidcore_dump;
80 char security_setidcore_path[MAXPATHLEN] = "/var/crash/%n.core";
81 uid_t security_setidcore_owner = 0;
82 gid_t security_setidcore_group = 0;
83 mode_t security_setidcore_mode = (S_IRUSR|S_IWUSR);
84 
85 static const u_int sysctl_flagmap[] = {
86 	PK_ADVLOCK, P_ADVLOCK,
87 	PK_EXEC, P_EXEC,
88 	PK_NOCLDWAIT, P_NOCLDWAIT,
89 	PK_32, P_32,
90 	PK_CLDSIGIGN, P_CLDSIGIGN,
91 	PK_SUGID, P_SUGID,
92 	0
93 };
94 
95 static const u_int sysctl_sflagmap[] = {
96 	PS_NOCLDSTOP, P_NOCLDSTOP,
97 	PS_WEXIT, P_WEXIT,
98 	PS_STOPFORK, P_STOPFORK,
99 	PS_STOPEXEC, P_STOPEXEC,
100 	PS_STOPEXIT, P_STOPEXIT,
101 	0
102 };
103 
104 static const u_int sysctl_slflagmap[] = {
105 	PSL_TRACED, P_TRACED,
106 	PSL_FSTRACE, P_FSTRACE,
107 	PSL_CHTRACED, P_CHTRACED,
108 	PSL_SYSCALL, P_SYSCALL,
109 	0
110 };
111 
112 static const u_int sysctl_lflagmap[] = {
113 	PL_CONTROLT, P_CONTROLT,
114 	PL_PPWAIT, P_PPWAIT,
115 	0
116 };
117 
118 static const u_int sysctl_stflagmap[] = {
119 	PST_PROFIL, P_PROFIL,
120 	0
121 
122 };
123 
124 static const u_int sysctl_lwpflagmap[] = {
125 	LW_INMEM, P_INMEM,
126 	LW_SINTR, P_SINTR,
127 	LW_SYSTEM, P_SYSTEM,
128 	0
129 };
130 
131 static const u_int sysctl_lwpprflagmap[] = {
132 	LPR_DETACHED, L_DETACHED,
133 	0
134 };
135 
136 /*
137  * try over estimating by 5 procs/lwps
138  */
139 #define KERN_PROCSLOP	(5 * sizeof(struct kinfo_proc))
140 #define KERN_LWPSLOP	(5 * sizeof(struct kinfo_lwp))
141 
142 static int dcopyout(struct lwp *, const void *, void *, size_t);
143 
144 static int
145 dcopyout(struct lwp *l, const void *kaddr, void *uaddr, size_t len)
146 {
147 	int error;
148 
149 	error = copyout(kaddr, uaddr, len);
150 	ktrmibio(-1, UIO_READ, uaddr, len, error);
151 
152 	return error;
153 }
154 
155 #ifdef DIAGNOSTIC
156 static int sysctl_kern_trigger_panic(SYSCTLFN_PROTO);
157 #endif
158 static int sysctl_kern_maxvnodes(SYSCTLFN_PROTO);
159 static int sysctl_kern_rtc_offset(SYSCTLFN_PROTO);
160 static int sysctl_kern_maxproc(SYSCTLFN_PROTO);
161 static int sysctl_kern_hostid(SYSCTLFN_PROTO);
162 static int sysctl_setlen(SYSCTLFN_PROTO);
163 static int sysctl_kern_clockrate(SYSCTLFN_PROTO);
164 static int sysctl_kern_file(SYSCTLFN_PROTO);
165 static int sysctl_msgbuf(SYSCTLFN_PROTO);
166 static int sysctl_kern_defcorename(SYSCTLFN_PROTO);
167 static int sysctl_kern_cptime(SYSCTLFN_PROTO);
168 #if NPTY > 0
169 static int sysctl_kern_maxptys(SYSCTLFN_PROTO);
170 #endif /* NPTY > 0 */
171 static int sysctl_kern_sbmax(SYSCTLFN_PROTO);
172 static int sysctl_kern_urnd(SYSCTLFN_PROTO);
173 static int sysctl_kern_arnd(SYSCTLFN_PROTO);
174 static int sysctl_kern_lwp(SYSCTLFN_PROTO);
175 static int sysctl_kern_forkfsleep(SYSCTLFN_PROTO);
176 static int sysctl_kern_root_partition(SYSCTLFN_PROTO);
177 static int sysctl_kern_drivers(SYSCTLFN_PROTO);
178 static int sysctl_kern_file2(SYSCTLFN_PROTO);
179 static int sysctl_security_setidcore(SYSCTLFN_PROTO);
180 static int sysctl_security_setidcorename(SYSCTLFN_PROTO);
181 static int sysctl_kern_cpid(SYSCTLFN_PROTO);
182 static int sysctl_doeproc(SYSCTLFN_PROTO);
183 static int sysctl_kern_proc_args(SYSCTLFN_PROTO);
184 static int sysctl_hw_usermem(SYSCTLFN_PROTO);
185 static int sysctl_hw_cnmagic(SYSCTLFN_PROTO);
186 
187 static u_int sysctl_map_flags(const u_int *, u_int);
188 static void fill_kproc2(struct proc *, struct kinfo_proc2 *, bool);
189 static void fill_lwp(struct lwp *l, struct kinfo_lwp *kl);
190 static void fill_file(struct kinfo_file *, const file_t *, const fdfile_t *,
191 		      int, pid_t);
192 
193 /*
194  * ********************************************************************
195  * section 1: setup routines
196  * ********************************************************************
197  * These functions are stuffed into a link set for sysctl setup
198  * functions. They're never called or referenced from anywhere else.
199  * ********************************************************************
200  */
201 
202 /*
203  * sets up the base nodes...
204  */
205 SYSCTL_SETUP(sysctl_root_setup, "sysctl base setup")
206 {
207 
208 	sysctl_createv(clog, 0, NULL, NULL,
209 		       CTLFLAG_PERMANENT,
210 		       CTLTYPE_NODE, "kern",
211 		       SYSCTL_DESCR("High kernel"),
212 		       NULL, 0, NULL, 0,
213 		       CTL_KERN, CTL_EOL);
214 	sysctl_createv(clog, 0, NULL, NULL,
215 		       CTLFLAG_PERMANENT,
216 		       CTLTYPE_NODE, "vm",
217 		       SYSCTL_DESCR("Virtual memory"),
218 		       NULL, 0, NULL, 0,
219 		       CTL_VM, CTL_EOL);
220 	sysctl_createv(clog, 0, NULL, NULL,
221 		       CTLFLAG_PERMANENT,
222 		       CTLTYPE_NODE, "vfs",
223 		       SYSCTL_DESCR("Filesystem"),
224 		       NULL, 0, NULL, 0,
225 		       CTL_VFS, CTL_EOL);
226 	sysctl_createv(clog, 0, NULL, NULL,
227 		       CTLFLAG_PERMANENT,
228 		       CTLTYPE_NODE, "net",
229 		       SYSCTL_DESCR("Networking"),
230 		       NULL, 0, NULL, 0,
231 		       CTL_NET, CTL_EOL);
232 	sysctl_createv(clog, 0, NULL, NULL,
233 		       CTLFLAG_PERMANENT,
234 		       CTLTYPE_NODE, "debug",
235 		       SYSCTL_DESCR("Debugging"),
236 		       NULL, 0, NULL, 0,
237 		       CTL_DEBUG, CTL_EOL);
238 	sysctl_createv(clog, 0, NULL, NULL,
239 		       CTLFLAG_PERMANENT,
240 		       CTLTYPE_NODE, "hw",
241 		       SYSCTL_DESCR("Generic CPU, I/O"),
242 		       NULL, 0, NULL, 0,
243 		       CTL_HW, CTL_EOL);
244 	sysctl_createv(clog, 0, NULL, NULL,
245 		       CTLFLAG_PERMANENT,
246 		       CTLTYPE_NODE, "machdep",
247 		       SYSCTL_DESCR("Machine dependent"),
248 		       NULL, 0, NULL, 0,
249 		       CTL_MACHDEP, CTL_EOL);
250 	/*
251 	 * this node is inserted so that the sysctl nodes in libc can
252 	 * operate.
253 	 */
254 	sysctl_createv(clog, 0, NULL, NULL,
255 		       CTLFLAG_PERMANENT,
256 		       CTLTYPE_NODE, "user",
257 		       SYSCTL_DESCR("User-level"),
258 		       NULL, 0, NULL, 0,
259 		       CTL_USER, CTL_EOL);
260 	sysctl_createv(clog, 0, NULL, NULL,
261 		       CTLFLAG_PERMANENT,
262 		       CTLTYPE_NODE, "ddb",
263 		       SYSCTL_DESCR("In-kernel debugger"),
264 		       NULL, 0, NULL, 0,
265 		       CTL_DDB, CTL_EOL);
266 	sysctl_createv(clog, 0, NULL, NULL,
267 		       CTLFLAG_PERMANENT,
268 		       CTLTYPE_NODE, "proc",
269 		       SYSCTL_DESCR("Per-process"),
270 		       NULL, 0, NULL, 0,
271 		       CTL_PROC, CTL_EOL);
272 	sysctl_createv(clog, 0, NULL, NULL,
273 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
274 		       CTLTYPE_NODE, "vendor",
275 		       SYSCTL_DESCR("Vendor specific"),
276 		       NULL, 0, NULL, 0,
277 		       CTL_VENDOR, CTL_EOL);
278 	sysctl_createv(clog, 0, NULL, NULL,
279 		       CTLFLAG_PERMANENT,
280 		       CTLTYPE_NODE, "emul",
281 		       SYSCTL_DESCR("Emulation settings"),
282 		       NULL, 0, NULL, 0,
283 		       CTL_EMUL, CTL_EOL);
284 	sysctl_createv(clog, 0, NULL, NULL,
285 		       CTLFLAG_PERMANENT,
286 		       CTLTYPE_NODE, "security",
287 		       SYSCTL_DESCR("Security"),
288 		       NULL, 0, NULL, 0,
289 		       CTL_SECURITY, CTL_EOL);
290 }
291 
292 /*
293  * this setup routine is a replacement for kern_sysctl()
294  */
295 SYSCTL_SETUP(sysctl_kern_setup, "sysctl kern subtree setup")
296 {
297 	extern int kern_logsigexit;	/* defined in kern/kern_sig.c */
298 	extern fixpt_t ccpu;		/* defined in kern/kern_synch.c */
299 	extern int dumponpanic;		/* defined in kern/subr_prf.c */
300 	const struct sysctlnode *rnode;
301 
302 	sysctl_createv(clog, 0, NULL, NULL,
303 		       CTLFLAG_PERMANENT,
304 		       CTLTYPE_NODE, "kern", NULL,
305 		       NULL, 0, NULL, 0,
306 		       CTL_KERN, CTL_EOL);
307 
308 	sysctl_createv(clog, 0, NULL, NULL,
309 		       CTLFLAG_PERMANENT,
310 		       CTLTYPE_STRING, "ostype",
311 		       SYSCTL_DESCR("Operating system type"),
312 		       NULL, 0, &ostype, 0,
313 		       CTL_KERN, KERN_OSTYPE, CTL_EOL);
314 	sysctl_createv(clog, 0, NULL, NULL,
315 		       CTLFLAG_PERMANENT,
316 		       CTLTYPE_STRING, "osrelease",
317 		       SYSCTL_DESCR("Operating system release"),
318 		       NULL, 0, &osrelease, 0,
319 		       CTL_KERN, KERN_OSRELEASE, CTL_EOL);
320 	sysctl_createv(clog, 0, NULL, NULL,
321 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
322 		       CTLTYPE_INT, "osrevision",
323 		       SYSCTL_DESCR("Operating system revision"),
324 		       NULL, __NetBSD_Version__, NULL, 0,
325 		       CTL_KERN, KERN_OSREV, CTL_EOL);
326 	sysctl_createv(clog, 0, NULL, NULL,
327 		       CTLFLAG_PERMANENT,
328 		       CTLTYPE_STRING, "version",
329 		       SYSCTL_DESCR("Kernel version"),
330 		       NULL, 0, &version, 0,
331 		       CTL_KERN, KERN_VERSION, CTL_EOL);
332 	sysctl_createv(clog, 0, NULL, NULL,
333 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
334 		       CTLTYPE_INT, "maxvnodes",
335 		       SYSCTL_DESCR("Maximum number of vnodes"),
336 		       sysctl_kern_maxvnodes, 0, NULL, 0,
337 		       CTL_KERN, KERN_MAXVNODES, CTL_EOL);
338 	sysctl_createv(clog, 0, NULL, NULL,
339 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
340 		       CTLTYPE_INT, "maxproc",
341 		       SYSCTL_DESCR("Maximum number of simultaneous processes"),
342 		       sysctl_kern_maxproc, 0, NULL, 0,
343 		       CTL_KERN, KERN_MAXPROC, CTL_EOL);
344 	sysctl_createv(clog, 0, NULL, NULL,
345 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
346 		       CTLTYPE_INT, "maxfiles",
347 		       SYSCTL_DESCR("Maximum number of open files"),
348 		       NULL, 0, &maxfiles, 0,
349 		       CTL_KERN, KERN_MAXFILES, CTL_EOL);
350 	sysctl_createv(clog, 0, NULL, NULL,
351 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
352 		       CTLTYPE_INT, "argmax",
353 		       SYSCTL_DESCR("Maximum number of bytes of arguments to "
354 				    "execve(2)"),
355 		       NULL, ARG_MAX, NULL, 0,
356 		       CTL_KERN, KERN_ARGMAX, CTL_EOL);
357 	sysctl_createv(clog, 0, NULL, NULL,
358 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
359 		       CTLTYPE_STRING, "hostname",
360 		       SYSCTL_DESCR("System hostname"),
361 		       sysctl_setlen, 0, &hostname, MAXHOSTNAMELEN,
362 		       CTL_KERN, KERN_HOSTNAME, CTL_EOL);
363 	sysctl_createv(clog, 0, NULL, NULL,
364 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE|CTLFLAG_HEX,
365 		       CTLTYPE_INT, "hostid",
366 		       SYSCTL_DESCR("System host ID number"),
367 		       sysctl_kern_hostid, 0, NULL, 0,
368 		       CTL_KERN, KERN_HOSTID, CTL_EOL);
369 	sysctl_createv(clog, 0, NULL, NULL,
370 		       CTLFLAG_PERMANENT,
371 		       CTLTYPE_STRUCT, "clockrate",
372 		       SYSCTL_DESCR("Kernel clock rates"),
373 		       sysctl_kern_clockrate, 0, NULL,
374 		       sizeof(struct clockinfo),
375 		       CTL_KERN, KERN_CLOCKRATE, CTL_EOL);
376 	sysctl_createv(clog, 0, NULL, NULL,
377 		       CTLFLAG_PERMANENT,
378 		       CTLTYPE_INT, "hardclock_ticks",
379 		       SYSCTL_DESCR("Number of hardclock ticks"),
380 		       NULL, 0, &hardclock_ticks, sizeof(hardclock_ticks),
381 		       CTL_KERN, KERN_HARDCLOCK_TICKS, CTL_EOL);
382 	sysctl_createv(clog, 0, NULL, NULL,
383 		       CTLFLAG_PERMANENT,
384 		       CTLTYPE_STRUCT, "vnode",
385 		       SYSCTL_DESCR("System vnode table"),
386 		       sysctl_kern_vnode, 0, NULL, 0,
387 		       CTL_KERN, KERN_VNODE, CTL_EOL);
388 	sysctl_createv(clog, 0, NULL, NULL,
389 		       CTLFLAG_PERMANENT,
390 		       CTLTYPE_STRUCT, "file",
391 		       SYSCTL_DESCR("System open file table"),
392 		       sysctl_kern_file, 0, NULL, 0,
393 		       CTL_KERN, KERN_FILE, CTL_EOL);
394 #ifndef GPROF
395 	sysctl_createv(clog, 0, NULL, NULL,
396 		       CTLFLAG_PERMANENT,
397 		       CTLTYPE_NODE, "profiling",
398 		       SYSCTL_DESCR("Profiling information (not available)"),
399 		       sysctl_notavail, 0, NULL, 0,
400 		       CTL_KERN, KERN_PROF, CTL_EOL);
401 #endif
402 	sysctl_createv(clog, 0, NULL, NULL,
403 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
404 		       CTLTYPE_INT, "posix1version",
405 		       SYSCTL_DESCR("Version of ISO/IEC 9945 (POSIX 1003.1) "
406 				    "with which the operating system attempts "
407 				    "to comply"),
408 		       NULL, _POSIX_VERSION, NULL, 0,
409 		       CTL_KERN, KERN_POSIX1, CTL_EOL);
410 	sysctl_createv(clog, 0, NULL, NULL,
411 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
412 		       CTLTYPE_INT, "ngroups",
413 		       SYSCTL_DESCR("Maximum number of supplemental groups"),
414 		       NULL, NGROUPS_MAX, NULL, 0,
415 		       CTL_KERN, KERN_NGROUPS, CTL_EOL);
416 	sysctl_createv(clog, 0, NULL, NULL,
417 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
418 		       CTLTYPE_INT, "job_control",
419 		       SYSCTL_DESCR("Whether job control is available"),
420 		       NULL, 1, NULL, 0,
421 		       CTL_KERN, KERN_JOB_CONTROL, CTL_EOL);
422 	sysctl_createv(clog, 0, NULL, NULL,
423 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
424 		       CTLTYPE_INT, "saved_ids",
425 		       SYSCTL_DESCR("Whether POSIX saved set-group/user ID is "
426 				    "available"), NULL,
427 #ifdef _POSIX_SAVED_IDS
428 		       1,
429 #else /* _POSIX_SAVED_IDS */
430 		       0,
431 #endif /* _POSIX_SAVED_IDS */
432 		       NULL, 0, CTL_KERN, KERN_SAVED_IDS, CTL_EOL);
433 	sysctl_createv(clog, 0, NULL, NULL,
434 		       CTLFLAG_PERMANENT,
435 		       CTLTYPE_STRUCT, "boottime",
436 		       SYSCTL_DESCR("System boot time"),
437 		       NULL, 0, &boottime, sizeof(boottime),
438 		       CTL_KERN, KERN_BOOTTIME, CTL_EOL);
439 	sysctl_createv(clog, 0, NULL, NULL,
440 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
441 		       CTLTYPE_STRING, "domainname",
442 		       SYSCTL_DESCR("YP domain name"),
443 		       sysctl_setlen, 0, &domainname, MAXHOSTNAMELEN,
444 		       CTL_KERN, KERN_DOMAINNAME, CTL_EOL);
445 	sysctl_createv(clog, 0, NULL, NULL,
446 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
447 		       CTLTYPE_INT, "maxpartitions",
448 		       SYSCTL_DESCR("Maximum number of partitions allowed per "
449 				    "disk"),
450 		       NULL, MAXPARTITIONS, NULL, 0,
451 		       CTL_KERN, KERN_MAXPARTITIONS, CTL_EOL);
452 	sysctl_createv(clog, 0, NULL, NULL,
453 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
454 		       CTLTYPE_INT, "rawpartition",
455 		       SYSCTL_DESCR("Raw partition of a disk"),
456 		       NULL, RAW_PART, NULL, 0,
457 		       CTL_KERN, KERN_RAWPARTITION, CTL_EOL);
458 	sysctl_createv(clog, 0, NULL, NULL,
459 		       CTLFLAG_PERMANENT,
460 		       CTLTYPE_STRUCT, "timex", NULL,
461 		       sysctl_notavail, 0, NULL, 0,
462 		       CTL_KERN, KERN_TIMEX, CTL_EOL);
463 	sysctl_createv(clog, 0, NULL, NULL,
464 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
465 		       CTLTYPE_INT, "rtc_offset",
466 		       SYSCTL_DESCR("Offset of real time clock from UTC in "
467 				    "minutes"),
468 		       sysctl_kern_rtc_offset, 0, &rtc_offset, 0,
469 		       CTL_KERN, KERN_RTC_OFFSET, CTL_EOL);
470 	sysctl_createv(clog, 0, NULL, NULL,
471 		       CTLFLAG_PERMANENT,
472 		       CTLTYPE_STRING, "root_device",
473 		       SYSCTL_DESCR("Name of the root device"),
474 		       sysctl_root_device, 0, NULL, 0,
475 		       CTL_KERN, KERN_ROOT_DEVICE, CTL_EOL);
476 	sysctl_createv(clog, 0, NULL, NULL,
477 		       CTLFLAG_PERMANENT,
478 		       CTLTYPE_INT, "msgbufsize",
479 		       SYSCTL_DESCR("Size of the kernel message buffer"),
480 		       sysctl_msgbuf, 0, NULL, 0,
481 		       CTL_KERN, KERN_MSGBUFSIZE, CTL_EOL);
482 	sysctl_createv(clog, 0, NULL, NULL,
483 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
484 		       CTLTYPE_INT, "fsync",
485 		       SYSCTL_DESCR("Whether the POSIX 1003.1b File "
486 				    "Synchronization Option is available on "
487 				    "this system"),
488 		       NULL, 1, NULL, 0,
489 		       CTL_KERN, KERN_FSYNC, CTL_EOL);
490 	sysctl_createv(clog, 0, NULL, NULL,
491 		       CTLFLAG_PERMANENT,
492 		       CTLTYPE_NODE, "ipc",
493 		       SYSCTL_DESCR("SysV IPC options"),
494 		       NULL, 0, NULL, 0,
495 		       CTL_KERN, KERN_SYSVIPC, CTL_EOL);
496 	sysctl_createv(clog, 0, NULL, NULL,
497 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
498 		       CTLTYPE_INT, "sysvmsg",
499 		       SYSCTL_DESCR("System V style message support available"),
500 		       NULL,
501 #ifdef SYSVMSG
502 		       1,
503 #else /* SYSVMSG */
504 		       0,
505 #endif /* SYSVMSG */
506 		       NULL, 0, CTL_KERN, KERN_SYSVIPC, KERN_SYSVIPC_MSG, CTL_EOL);
507 	sysctl_createv(clog, 0, NULL, NULL,
508 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
509 		       CTLTYPE_INT, "sysvsem",
510 		       SYSCTL_DESCR("System V style semaphore support "
511 				    "available"), NULL,
512 #ifdef SYSVSEM
513 		       1,
514 #else /* SYSVSEM */
515 		       0,
516 #endif /* SYSVSEM */
517 		       NULL, 0, CTL_KERN, KERN_SYSVIPC, KERN_SYSVIPC_SEM, CTL_EOL);
518 	sysctl_createv(clog, 0, NULL, NULL,
519 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
520 		       CTLTYPE_INT, "sysvshm",
521 		       SYSCTL_DESCR("System V style shared memory support "
522 				    "available"), NULL,
523 #ifdef SYSVSHM
524 		       1,
525 #else /* SYSVSHM */
526 		       0,
527 #endif /* SYSVSHM */
528 		       NULL, 0, CTL_KERN, KERN_SYSVIPC, KERN_SYSVIPC_SHM, CTL_EOL);
529 	sysctl_createv(clog, 0, NULL, NULL,
530 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
531 		       CTLTYPE_INT, "synchronized_io",
532 		       SYSCTL_DESCR("Whether the POSIX 1003.1b Synchronized "
533 				    "I/O Option is available on this system"),
534 		       NULL, 1, NULL, 0,
535 		       CTL_KERN, KERN_SYNCHRONIZED_IO, CTL_EOL);
536 	sysctl_createv(clog, 0, NULL, NULL,
537 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
538 		       CTLTYPE_INT, "iov_max",
539 		       SYSCTL_DESCR("Maximum number of iovec structures per "
540 				    "process"),
541 		       NULL, IOV_MAX, NULL, 0,
542 		       CTL_KERN, KERN_IOV_MAX, CTL_EOL);
543 	sysctl_createv(clog, 0, NULL, NULL,
544 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
545 		       CTLTYPE_INT, "mapped_files",
546 		       SYSCTL_DESCR("Whether the POSIX 1003.1b Memory Mapped "
547 				    "Files Option is available on this system"),
548 		       NULL, 1, NULL, 0,
549 		       CTL_KERN, KERN_MAPPED_FILES, CTL_EOL);
550 	sysctl_createv(clog, 0, NULL, NULL,
551 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
552 		       CTLTYPE_INT, "memlock",
553 		       SYSCTL_DESCR("Whether the POSIX 1003.1b Process Memory "
554 				    "Locking Option is available on this "
555 				    "system"),
556 		       NULL, 1, NULL, 0,
557 		       CTL_KERN, KERN_MEMLOCK, CTL_EOL);
558 	sysctl_createv(clog, 0, NULL, NULL,
559 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
560 		       CTLTYPE_INT, "memlock_range",
561 		       SYSCTL_DESCR("Whether the POSIX 1003.1b Range Memory "
562 				    "Locking Option is available on this "
563 				    "system"),
564 		       NULL, 1, NULL, 0,
565 		       CTL_KERN, KERN_MEMLOCK_RANGE, CTL_EOL);
566 	sysctl_createv(clog, 0, NULL, NULL,
567 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
568 		       CTLTYPE_INT, "memory_protection",
569 		       SYSCTL_DESCR("Whether the POSIX 1003.1b Memory "
570 				    "Protection Option is available on this "
571 				    "system"),
572 		       NULL, 1, NULL, 0,
573 		       CTL_KERN, KERN_MEMORY_PROTECTION, CTL_EOL);
574 	sysctl_createv(clog, 0, NULL, NULL,
575 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
576 		       CTLTYPE_INT, "login_name_max",
577 		       SYSCTL_DESCR("Maximum login name length"),
578 		       NULL, LOGIN_NAME_MAX, NULL, 0,
579 		       CTL_KERN, KERN_LOGIN_NAME_MAX, CTL_EOL);
580 	sysctl_createv(clog, 0, NULL, NULL,
581 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
582 		       CTLTYPE_STRING, "defcorename",
583 		       SYSCTL_DESCR("Default core file name"),
584 		       sysctl_kern_defcorename, 0, defcorename, MAXPATHLEN,
585 		       CTL_KERN, KERN_DEFCORENAME, CTL_EOL);
586 	sysctl_createv(clog, 0, NULL, NULL,
587 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
588 		       CTLTYPE_INT, "logsigexit",
589 		       SYSCTL_DESCR("Log process exit when caused by signals"),
590 		       NULL, 0, &kern_logsigexit, 0,
591 		       CTL_KERN, KERN_LOGSIGEXIT, CTL_EOL);
592 	sysctl_createv(clog, 0, NULL, NULL,
593 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
594 		       CTLTYPE_INT, "fscale",
595 		       SYSCTL_DESCR("Kernel fixed-point scale factor"),
596 		       NULL, FSCALE, NULL, 0,
597 		       CTL_KERN, KERN_FSCALE, CTL_EOL);
598 	sysctl_createv(clog, 0, NULL, NULL,
599 		       CTLFLAG_PERMANENT,
600 		       CTLTYPE_INT, "ccpu",
601 		       SYSCTL_DESCR("Scheduler exponential decay value"),
602 		       NULL, 0, &ccpu, 0,
603 		       CTL_KERN, KERN_CCPU, CTL_EOL);
604 	sysctl_createv(clog, 0, NULL, NULL,
605 		       CTLFLAG_PERMANENT,
606 		       CTLTYPE_STRUCT, "cp_time",
607 		       SYSCTL_DESCR("Clock ticks spent in different CPU states"),
608 		       sysctl_kern_cptime, 0, NULL, 0,
609 		       CTL_KERN, KERN_CP_TIME, CTL_EOL);
610 	sysctl_createv(clog, 0, NULL, NULL,
611 		       CTLFLAG_PERMANENT,
612 		       CTLTYPE_INT, "msgbuf",
613 		       SYSCTL_DESCR("Kernel message buffer"),
614 		       sysctl_msgbuf, 0, NULL, 0,
615 		       CTL_KERN, KERN_MSGBUF, CTL_EOL);
616 	sysctl_createv(clog, 0, NULL, NULL,
617 		       CTLFLAG_PERMANENT,
618 		       CTLTYPE_STRUCT, "consdev",
619 		       SYSCTL_DESCR("Console device"),
620 		       sysctl_consdev, 0, NULL, sizeof(dev_t),
621 		       CTL_KERN, KERN_CONSDEV, CTL_EOL);
622 #if NPTY > 0
623 	sysctl_createv(clog, 0, NULL, NULL,
624 		       CTLFLAG_PERMANENT,
625 		       CTLTYPE_INT, "maxptys",
626 		       SYSCTL_DESCR("Maximum number of pseudo-ttys"),
627 		       sysctl_kern_maxptys, 0, NULL, 0,
628 		       CTL_KERN, KERN_MAXPTYS, CTL_EOL);
629 #endif /* NPTY > 0 */
630 	sysctl_createv(clog, 0, NULL, NULL,
631 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
632 		       CTLTYPE_INT, "maxphys",
633 		       SYSCTL_DESCR("Maximum raw I/O transfer size"),
634 		       NULL, MAXPHYS, NULL, 0,
635 		       CTL_KERN, KERN_MAXPHYS, CTL_EOL);
636 	sysctl_createv(clog, 0, NULL, NULL,
637 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
638 		       CTLTYPE_INT, "sbmax",
639 		       SYSCTL_DESCR("Maximum socket buffer size"),
640 		       sysctl_kern_sbmax, 0, NULL, 0,
641 		       CTL_KERN, KERN_SBMAX, CTL_EOL);
642 	sysctl_createv(clog, 0, NULL, NULL,
643 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
644 		       CTLTYPE_INT, "monotonic_clock",
645 		       SYSCTL_DESCR("Implementation version of the POSIX "
646 				    "1003.1b Monotonic Clock Option"),
647 		       /* XXX _POSIX_VERSION */
648 		       NULL, _POSIX_MONOTONIC_CLOCK, NULL, 0,
649 		       CTL_KERN, KERN_MONOTONIC_CLOCK, CTL_EOL);
650 	sysctl_createv(clog, 0, NULL, NULL,
651 		       CTLFLAG_PERMANENT,
652 		       CTLTYPE_INT, "urandom",
653 		       SYSCTL_DESCR("Random integer value"),
654 		       sysctl_kern_urnd, 0, NULL, 0,
655 		       CTL_KERN, KERN_URND, CTL_EOL);
656 	sysctl_createv(clog, 0, NULL, NULL,
657 		       CTLFLAG_PERMANENT,
658 		       CTLTYPE_INT, "arandom",
659 		       SYSCTL_DESCR("n bytes of random data"),
660 		       sysctl_kern_arnd, 0, NULL, 0,
661 		       CTL_KERN, KERN_ARND, CTL_EOL);
662 	sysctl_createv(clog, 0, NULL, NULL,
663 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
664 		       CTLTYPE_INT, "labelsector",
665 		       SYSCTL_DESCR("Sector number containing the disklabel"),
666 		       NULL, LABELSECTOR, NULL, 0,
667 		       CTL_KERN, KERN_LABELSECTOR, CTL_EOL);
668 	sysctl_createv(clog, 0, NULL, NULL,
669 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
670 		       CTLTYPE_INT, "labeloffset",
671 		       SYSCTL_DESCR("Offset of the disklabel within the "
672 				    "sector"),
673 		       NULL, LABELOFFSET, NULL, 0,
674 		       CTL_KERN, KERN_LABELOFFSET, CTL_EOL);
675 	sysctl_createv(clog, 0, NULL, NULL,
676 		       CTLFLAG_PERMANENT,
677 		       CTLTYPE_NODE, "lwp",
678 		       SYSCTL_DESCR("System-wide LWP information"),
679 		       sysctl_kern_lwp, 0, NULL, 0,
680 		       CTL_KERN, KERN_LWP, CTL_EOL);
681 	sysctl_createv(clog, 0, NULL, NULL,
682 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
683 		       CTLTYPE_INT, "forkfsleep",
684 		       SYSCTL_DESCR("Milliseconds to sleep on fork failure due "
685 				    "to process limits"),
686 		       sysctl_kern_forkfsleep, 0, NULL, 0,
687 		       CTL_KERN, KERN_FORKFSLEEP, CTL_EOL);
688 	sysctl_createv(clog, 0, NULL, NULL,
689 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
690 		       CTLTYPE_INT, "posix_threads",
691 		       SYSCTL_DESCR("Version of IEEE Std 1003.1 and its "
692 				    "Threads option to which the system "
693 				    "attempts to conform"),
694 		       /* XXX _POSIX_VERSION */
695 		       NULL, _POSIX_THREADS, NULL, 0,
696 		       CTL_KERN, KERN_POSIX_THREADS, CTL_EOL);
697 	sysctl_createv(clog, 0, NULL, NULL,
698 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
699 		       CTLTYPE_INT, "posix_semaphores",
700 		       SYSCTL_DESCR("Version of IEEE Std 1003.1 and its "
701 				    "Semaphores option to which the system "
702 				    "attempts to conform"), NULL,
703 #ifdef P1003_1B_SEMAPHORE
704 		       200112,
705 #else /* P1003_1B_SEMAPHORE */
706 		       0,
707 #endif /* P1003_1B_SEMAPHORE */
708 		       NULL, 0, CTL_KERN, KERN_POSIX_SEMAPHORES, CTL_EOL);
709 	sysctl_createv(clog, 0, NULL, NULL,
710 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
711 		       CTLTYPE_INT, "posix_barriers",
712 		       SYSCTL_DESCR("Version of IEEE Std 1003.1 and its "
713 				    "Barriers option to which the system "
714 				    "attempts to conform"),
715 		       /* XXX _POSIX_VERSION */
716 		       NULL, _POSIX_BARRIERS, NULL, 0,
717 		       CTL_KERN, KERN_POSIX_BARRIERS, CTL_EOL);
718 	sysctl_createv(clog, 0, NULL, NULL,
719 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
720 		       CTLTYPE_INT, "posix_timers",
721 		       SYSCTL_DESCR("Version of IEEE Std 1003.1 and its "
722 				    "Timers option to which the system "
723 				    "attempts to conform"),
724 		       /* XXX _POSIX_VERSION */
725 		       NULL, _POSIX_TIMERS, NULL, 0,
726 		       CTL_KERN, KERN_POSIX_TIMERS, CTL_EOL);
727 	sysctl_createv(clog, 0, NULL, NULL,
728 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
729 		       CTLTYPE_INT, "posix_spin_locks",
730 		       SYSCTL_DESCR("Version of IEEE Std 1003.1 and its Spin "
731 				    "Locks option to which the system attempts "
732 				    "to conform"),
733 		       /* XXX _POSIX_VERSION */
734 		       NULL, _POSIX_SPIN_LOCKS, NULL, 0,
735 		       CTL_KERN, KERN_POSIX_SPIN_LOCKS, CTL_EOL);
736 	sysctl_createv(clog, 0, NULL, NULL,
737 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
738 		       CTLTYPE_INT, "posix_reader_writer_locks",
739 		       SYSCTL_DESCR("Version of IEEE Std 1003.1 and its "
740 				    "Read-Write Locks option to which the "
741 				    "system attempts to conform"),
742 		       /* XXX _POSIX_VERSION */
743 		       NULL, _POSIX_READER_WRITER_LOCKS, NULL, 0,
744 		       CTL_KERN, KERN_POSIX_READER_WRITER_LOCKS, CTL_EOL);
745 	sysctl_createv(clog, 0, NULL, NULL,
746 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
747 		       CTLTYPE_INT, "dump_on_panic",
748 		       SYSCTL_DESCR("Perform a crash dump on system panic"),
749 		       NULL, 0, &dumponpanic, 0,
750 		       CTL_KERN, KERN_DUMP_ON_PANIC, CTL_EOL);
751 #ifdef DIAGNOSTIC
752 	sysctl_createv(clog, 0, NULL, NULL,
753 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
754 		       CTLTYPE_INT, "panic_now",
755 		       SYSCTL_DESCR("Trigger a panic"),
756 		       sysctl_kern_trigger_panic, 0, NULL, 0,
757 		       CTL_KERN, CTL_CREATE, CTL_EOL);
758 #endif
759 	sysctl_createv(clog, 0, NULL, NULL,
760 		       CTLFLAG_PERMANENT,
761 		       CTLTYPE_INT, "root_partition",
762 		       SYSCTL_DESCR("Root partition on the root device"),
763 		       sysctl_kern_root_partition, 0, NULL, 0,
764 		       CTL_KERN, KERN_ROOT_PARTITION, CTL_EOL);
765 	sysctl_createv(clog, 0, NULL, NULL,
766 		       CTLFLAG_PERMANENT,
767 		       CTLTYPE_STRUCT, "drivers",
768 		       SYSCTL_DESCR("List of all drivers with block and "
769 				    "character device numbers"),
770 		       sysctl_kern_drivers, 0, NULL, 0,
771 		       CTL_KERN, KERN_DRIVERS, CTL_EOL);
772 	sysctl_createv(clog, 0, NULL, NULL,
773 		       CTLFLAG_PERMANENT,
774 		       CTLTYPE_STRUCT, "file2",
775 		       SYSCTL_DESCR("System open file table"),
776 		       sysctl_kern_file2, 0, NULL, 0,
777 		       CTL_KERN, KERN_FILE2, CTL_EOL);
778 	sysctl_createv(clog, 0, NULL, NULL,
779 		       CTLFLAG_PERMANENT,
780 		       CTLTYPE_STRUCT, "cp_id",
781 		       SYSCTL_DESCR("Mapping of CPU number to CPU id"),
782 		       sysctl_kern_cpid, 0, NULL, 0,
783 		       CTL_KERN, KERN_CP_ID, CTL_EOL);
784 	sysctl_createv(clog, 0, NULL, &rnode,
785 		       CTLFLAG_PERMANENT,
786 		       CTLTYPE_NODE, "coredump",
787 		       SYSCTL_DESCR("Coredump settings."),
788 		       NULL, 0, NULL, 0,
789 		       CTL_KERN, CTL_CREATE, CTL_EOL);
790 	sysctl_createv(clog, 0, &rnode, &rnode,
791 		       CTLFLAG_PERMANENT,
792 		       CTLTYPE_NODE, "setid",
793 		       SYSCTL_DESCR("Set-id processes' coredump settings."),
794 		       NULL, 0, NULL, 0,
795 		       CTL_CREATE, CTL_EOL);
796 	sysctl_createv(clog, 0, &rnode, NULL,
797 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
798 		       CTLTYPE_INT, "dump",
799 		       SYSCTL_DESCR("Allow set-id processes to dump core."),
800 		       sysctl_security_setidcore, 0, &security_setidcore_dump,
801 		       sizeof(security_setidcore_dump),
802 		       CTL_CREATE, CTL_EOL);
803 	sysctl_createv(clog, 0, &rnode, NULL,
804 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
805 		       CTLTYPE_STRING, "path",
806 		       SYSCTL_DESCR("Path pattern for set-id coredumps."),
807 		       sysctl_security_setidcorename, 0,
808 		       &security_setidcore_path,
809 		       sizeof(security_setidcore_path),
810 		       CTL_CREATE, CTL_EOL);
811 	sysctl_createv(clog, 0, &rnode, NULL,
812 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
813 		       CTLTYPE_INT, "owner",
814 		       SYSCTL_DESCR("Owner id for set-id processes' cores."),
815 		       sysctl_security_setidcore, 0, &security_setidcore_owner,
816 		       0,
817 		       CTL_CREATE, CTL_EOL);
818 	sysctl_createv(clog, 0, &rnode, NULL,
819 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
820 		       CTLTYPE_INT, "group",
821 		       SYSCTL_DESCR("Group id for set-id processes' cores."),
822 		       sysctl_security_setidcore, 0, &security_setidcore_group,
823 		       0,
824 		       CTL_CREATE, CTL_EOL);
825 	sysctl_createv(clog, 0, &rnode, NULL,
826 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
827 		       CTLTYPE_INT, "mode",
828 		       SYSCTL_DESCR("Mode for set-id processes' cores."),
829 		       sysctl_security_setidcore, 0, &security_setidcore_mode,
830 		       0,
831 		       CTL_CREATE, CTL_EOL);
832 	sysctl_createv(clog, 0, NULL, NULL,
833 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
834 		       CTLTYPE_INT, "no_sa_support",
835 		       SYSCTL_DESCR("0 if the kernel supports SA, otherwise it doesn't"),
836 		       NULL, 1, NULL, 0,
837 		       CTL_KERN, CTL_CREATE, CTL_EOL);
838 }
839 
840 SYSCTL_SETUP(sysctl_kern_proc_setup,
841 	     "sysctl kern.proc/proc2/proc_args subtree setup")
842 {
843 
844 	sysctl_createv(clog, 0, NULL, NULL,
845 		       CTLFLAG_PERMANENT,
846 		       CTLTYPE_NODE, "kern", NULL,
847 		       NULL, 0, NULL, 0,
848 		       CTL_KERN, CTL_EOL);
849 
850 	sysctl_createv(clog, 0, NULL, NULL,
851 		       CTLFLAG_PERMANENT,
852 		       CTLTYPE_NODE, "proc",
853 		       SYSCTL_DESCR("System-wide process information"),
854 		       sysctl_doeproc, 0, NULL, 0,
855 		       CTL_KERN, KERN_PROC, CTL_EOL);
856 	sysctl_createv(clog, 0, NULL, NULL,
857 		       CTLFLAG_PERMANENT,
858 		       CTLTYPE_NODE, "proc2",
859 		       SYSCTL_DESCR("Machine-independent process information"),
860 		       sysctl_doeproc, 0, NULL, 0,
861 		       CTL_KERN, KERN_PROC2, CTL_EOL);
862 	sysctl_createv(clog, 0, NULL, NULL,
863 		       CTLFLAG_PERMANENT,
864 		       CTLTYPE_NODE, "proc_args",
865 		       SYSCTL_DESCR("Process argument information"),
866 		       sysctl_kern_proc_args, 0, NULL, 0,
867 		       CTL_KERN, KERN_PROC_ARGS, CTL_EOL);
868 
869 	/*
870 	  "nodes" under these:
871 
872 	  KERN_PROC_ALL
873 	  KERN_PROC_PID pid
874 	  KERN_PROC_PGRP pgrp
875 	  KERN_PROC_SESSION sess
876 	  KERN_PROC_TTY tty
877 	  KERN_PROC_UID uid
878 	  KERN_PROC_RUID uid
879 	  KERN_PROC_GID gid
880 	  KERN_PROC_RGID gid
881 
882 	  all in all, probably not worth the effort...
883 	*/
884 }
885 
886 SYSCTL_SETUP(sysctl_hw_setup, "sysctl hw subtree setup")
887 {
888 	u_int u;
889 	u_quad_t q;
890 
891 	sysctl_createv(clog, 0, NULL, NULL,
892 		       CTLFLAG_PERMANENT,
893 		       CTLTYPE_NODE, "hw", NULL,
894 		       NULL, 0, NULL, 0,
895 		       CTL_HW, CTL_EOL);
896 
897 	sysctl_createv(clog, 0, NULL, NULL,
898 		       CTLFLAG_PERMANENT,
899 		       CTLTYPE_STRING, "machine",
900 		       SYSCTL_DESCR("Machine class"),
901 		       NULL, 0, machine, 0,
902 		       CTL_HW, HW_MACHINE, CTL_EOL);
903 	sysctl_createv(clog, 0, NULL, NULL,
904 		       CTLFLAG_PERMANENT,
905 		       CTLTYPE_STRING, "model",
906 		       SYSCTL_DESCR("Machine model"),
907 		       NULL, 0, cpu_model, 0,
908 		       CTL_HW, HW_MODEL, CTL_EOL);
909 	sysctl_createv(clog, 0, NULL, NULL,
910 		       CTLFLAG_PERMANENT,
911 		       CTLTYPE_INT, "ncpu",
912 		       SYSCTL_DESCR("Number of CPUs configured"),
913 		       NULL, 0, &ncpu, 0,
914 		       CTL_HW, HW_NCPU, CTL_EOL);
915 	sysctl_createv(clog, 0, NULL, NULL,
916 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
917 		       CTLTYPE_INT, "byteorder",
918 		       SYSCTL_DESCR("System byte order"),
919 		       NULL, BYTE_ORDER, NULL, 0,
920 		       CTL_HW, HW_BYTEORDER, CTL_EOL);
921 	u = ((u_int)physmem > (UINT_MAX / PAGE_SIZE)) ?
922 		UINT_MAX : physmem * PAGE_SIZE;
923 	sysctl_createv(clog, 0, NULL, NULL,
924 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
925 		       CTLTYPE_INT, "physmem",
926 		       SYSCTL_DESCR("Bytes of physical memory"),
927 		       NULL, u, NULL, 0,
928 		       CTL_HW, HW_PHYSMEM, CTL_EOL);
929 	sysctl_createv(clog, 0, NULL, NULL,
930 		       CTLFLAG_PERMANENT,
931 		       CTLTYPE_INT, "usermem",
932 		       SYSCTL_DESCR("Bytes of non-kernel memory"),
933 		       sysctl_hw_usermem, 0, NULL, 0,
934 		       CTL_HW, HW_USERMEM, CTL_EOL);
935 	sysctl_createv(clog, 0, NULL, NULL,
936 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
937 		       CTLTYPE_INT, "pagesize",
938 		       SYSCTL_DESCR("Software page size"),
939 		       NULL, PAGE_SIZE, NULL, 0,
940 		       CTL_HW, HW_PAGESIZE, CTL_EOL);
941 	sysctl_createv(clog, 0, NULL, NULL,
942 		       CTLFLAG_PERMANENT,
943 		       CTLTYPE_STRING, "machine_arch",
944 		       SYSCTL_DESCR("Machine CPU class"),
945 		       NULL, 0, machine_arch, 0,
946 		       CTL_HW, HW_MACHINE_ARCH, CTL_EOL);
947 	sysctl_createv(clog, 0, NULL, NULL,
948 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
949 		       CTLTYPE_INT, "alignbytes",
950 		       SYSCTL_DESCR("Alignment constraint for all possible "
951 				    "data types"),
952 		       NULL, ALIGNBYTES, NULL, 0,
953 		       CTL_HW, HW_ALIGNBYTES, CTL_EOL);
954 	sysctl_createv(clog, 0, NULL, NULL,
955 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE|CTLFLAG_HEX,
956 		       CTLTYPE_STRING, "cnmagic",
957 		       SYSCTL_DESCR("Console magic key sequence"),
958 		       sysctl_hw_cnmagic, 0, NULL, CNS_LEN,
959 		       CTL_HW, HW_CNMAGIC, CTL_EOL);
960 	q = (u_quad_t)physmem * PAGE_SIZE;
961 	sysctl_createv(clog, 0, NULL, NULL,
962 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
963 		       CTLTYPE_QUAD, "physmem64",
964 		       SYSCTL_DESCR("Bytes of physical memory"),
965 		       NULL, q, NULL, 0,
966 		       CTL_HW, HW_PHYSMEM64, CTL_EOL);
967 	sysctl_createv(clog, 0, NULL, NULL,
968 		       CTLFLAG_PERMANENT,
969 		       CTLTYPE_QUAD, "usermem64",
970 		       SYSCTL_DESCR("Bytes of non-kernel memory"),
971 		       sysctl_hw_usermem, 0, NULL, 0,
972 		       CTL_HW, HW_USERMEM64, CTL_EOL);
973 	sysctl_createv(clog, 0, NULL, NULL,
974 		       CTLFLAG_PERMANENT,
975 		       CTLTYPE_INT, "ncpuonline",
976 		       SYSCTL_DESCR("Number of CPUs online"),
977 		       NULL, 0, &ncpuonline, 0,
978 		       CTL_HW, HW_NCPUONLINE, CTL_EOL);
979 }
980 
981 #ifdef DEBUG
982 /*
983  * Debugging related system variables.
984  */
985 struct ctldebug /* debug0, */ /* debug1, */ debug2, debug3, debug4;
986 struct ctldebug debug5, debug6, debug7, debug8, debug9;
987 struct ctldebug debug10, debug11, debug12, debug13, debug14;
988 struct ctldebug debug15, debug16, debug17, debug18, debug19;
989 static struct ctldebug *debugvars[CTL_DEBUG_MAXID] = {
990 	&debug0, &debug1, &debug2, &debug3, &debug4,
991 	&debug5, &debug6, &debug7, &debug8, &debug9,
992 	&debug10, &debug11, &debug12, &debug13, &debug14,
993 	&debug15, &debug16, &debug17, &debug18, &debug19,
994 };
995 
996 /*
997  * this setup routine is a replacement for debug_sysctl()
998  *
999  * note that it creates several nodes per defined debug variable
1000  */
1001 SYSCTL_SETUP(sysctl_debug_setup, "sysctl debug subtree setup")
1002 {
1003 	struct ctldebug *cdp;
1004 	char nodename[20];
1005 	int i;
1006 
1007 	/*
1008 	 * two ways here:
1009 	 *
1010 	 * the "old" way (debug.name -> value) which was emulated by
1011 	 * the sysctl(8) binary
1012 	 *
1013 	 * the new way, which the sysctl(8) binary was actually using
1014 
1015 	 node	debug
1016 	 node	debug.0
1017 	 string debug.0.name
1018 	 int	debug.0.value
1019 	 int	debug.name
1020 
1021 	 */
1022 
1023 	sysctl_createv(clog, 0, NULL, NULL,
1024 		       CTLFLAG_PERMANENT,
1025 		       CTLTYPE_NODE, "debug", NULL,
1026 		       NULL, 0, NULL, 0,
1027 		       CTL_DEBUG, CTL_EOL);
1028 
1029 	for (i = 0; i < CTL_DEBUG_MAXID; i++) {
1030 		cdp = debugvars[i];
1031 		if (cdp->debugname == NULL || cdp->debugvar == NULL)
1032 			continue;
1033 
1034 		snprintf(nodename, sizeof(nodename), "debug%d", i);
1035 		sysctl_createv(clog, 0, NULL, NULL,
1036 			       CTLFLAG_PERMANENT|CTLFLAG_HIDDEN,
1037 			       CTLTYPE_NODE, nodename, NULL,
1038 			       NULL, 0, NULL, 0,
1039 			       CTL_DEBUG, i, CTL_EOL);
1040 		sysctl_createv(clog, 0, NULL, NULL,
1041 			       CTLFLAG_PERMANENT|CTLFLAG_HIDDEN,
1042 			       CTLTYPE_STRING, "name", NULL,
1043 			       /*XXXUNCONST*/
1044 			       NULL, 0, __UNCONST(cdp->debugname), 0,
1045 			       CTL_DEBUG, i, CTL_DEBUG_NAME, CTL_EOL);
1046 		sysctl_createv(clog, 0, NULL, NULL,
1047 			       CTLFLAG_PERMANENT|CTLFLAG_HIDDEN,
1048 			       CTLTYPE_INT, "value", NULL,
1049 			       NULL, 0, cdp->debugvar, 0,
1050 			       CTL_DEBUG, i, CTL_DEBUG_VALUE, CTL_EOL);
1051 		sysctl_createv(clog, 0, NULL, NULL,
1052 			       CTLFLAG_PERMANENT,
1053 			       CTLTYPE_INT, cdp->debugname, NULL,
1054 			       NULL, 0, cdp->debugvar, 0,
1055 			       CTL_DEBUG, CTL_CREATE, CTL_EOL);
1056 	}
1057 }
1058 #endif /* DEBUG */
1059 
1060 /*
1061  * ********************************************************************
1062  * section 2: private node-specific helper routines.
1063  * ********************************************************************
1064  */
1065 
1066 #ifdef DIAGNOSTIC
1067 static int
1068 sysctl_kern_trigger_panic(SYSCTLFN_ARGS)
1069 {
1070 	int newtrig, error;
1071 	struct sysctlnode node;
1072 
1073 	newtrig = 0;
1074 	node = *rnode;
1075 	node.sysctl_data = &newtrig;
1076 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1077 	if (error || newp == NULL)
1078 		return (error);
1079 
1080 	if (newtrig != 0)
1081 		panic("Panic triggered");
1082 
1083 	return (error);
1084 }
1085 #endif
1086 
1087 /*
1088  * sysctl helper routine for kern.maxvnodes.  Drain vnodes if
1089  * new value is lower than desiredvnodes and then calls reinit
1090  * routines that needs to adjust to the new value.
1091  */
1092 static int
1093 sysctl_kern_maxvnodes(SYSCTLFN_ARGS)
1094 {
1095 	int error, new_vnodes, old_vnodes, new_max;
1096 	struct sysctlnode node;
1097 
1098 	new_vnodes = desiredvnodes;
1099 	node = *rnode;
1100 	node.sysctl_data = &new_vnodes;
1101 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1102 	if (error || newp == NULL)
1103 		return (error);
1104 
1105 	/* Limits: 75% of KVA and physical memory. */
1106 	new_max = calc_cache_size(kernel_map, 75, 75) / VNODE_COST;
1107 	if (new_vnodes > new_max)
1108 		new_vnodes = new_max;
1109 
1110 	old_vnodes = desiredvnodes;
1111 	desiredvnodes = new_vnodes;
1112 	if (new_vnodes < old_vnodes) {
1113 		error = vfs_drainvnodes(new_vnodes, l);
1114 		if (error) {
1115 			desiredvnodes = old_vnodes;
1116 			return (error);
1117 		}
1118 	}
1119 	vfs_reinit();
1120 	nchreinit();
1121 
1122 	return (0);
1123 }
1124 
1125 /*
1126  * sysctl helper routine for rtc_offset - set time after changes
1127  */
1128 static int
1129 sysctl_kern_rtc_offset(SYSCTLFN_ARGS)
1130 {
1131 	struct timespec ts, delta;
1132 	int error, new_rtc_offset;
1133 	struct sysctlnode node;
1134 
1135 	new_rtc_offset = rtc_offset;
1136 	node = *rnode;
1137 	node.sysctl_data = &new_rtc_offset;
1138 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1139 	if (error || newp == NULL)
1140 		return (error);
1141 
1142 	if (kauth_authorize_system(l->l_cred, KAUTH_SYSTEM_TIME,
1143 	    KAUTH_REQ_SYSTEM_TIME_RTCOFFSET,
1144 	    KAUTH_ARG(new_rtc_offset), NULL, NULL))
1145 		return (EPERM);
1146 	if (rtc_offset == new_rtc_offset)
1147 		return (0);
1148 
1149 	/* if we change the offset, adjust the time */
1150 	nanotime(&ts);
1151 	delta.tv_sec = 60 * (new_rtc_offset - rtc_offset);
1152 	delta.tv_nsec = 0;
1153 	timespecadd(&ts, &delta, &ts);
1154 	rtc_offset = new_rtc_offset;
1155 	return (settime(l->l_proc, &ts));
1156 }
1157 
1158 /*
1159  * sysctl helper routine for kern.maxproc. Ensures that the new
1160  * values are not too low or too high.
1161  */
1162 static int
1163 sysctl_kern_maxproc(SYSCTLFN_ARGS)
1164 {
1165 	int error, nmaxproc;
1166 	struct sysctlnode node;
1167 
1168 	nmaxproc = maxproc;
1169 	node = *rnode;
1170 	node.sysctl_data = &nmaxproc;
1171 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1172 	if (error || newp == NULL)
1173 		return (error);
1174 
1175 	if (nmaxproc < 0 || nmaxproc >= PID_MAX)
1176 		return (EINVAL);
1177 #ifdef __HAVE_CPU_MAXPROC
1178 	if (nmaxproc > cpu_maxproc())
1179 		return (EINVAL);
1180 #endif
1181 	maxproc = nmaxproc;
1182 
1183 	return (0);
1184 }
1185 
1186 /*
1187  * sysctl helper function for kern.hostid. The hostid is a long, but
1188  * we export it as an int, so we need to give it a little help.
1189  */
1190 static int
1191 sysctl_kern_hostid(SYSCTLFN_ARGS)
1192 {
1193 	int error, inthostid;
1194 	struct sysctlnode node;
1195 
1196 	inthostid = hostid;  /* XXX assumes sizeof int <= sizeof long */
1197 	node = *rnode;
1198 	node.sysctl_data = &inthostid;
1199 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1200 	if (error || newp == NULL)
1201 		return (error);
1202 
1203 	hostid = (unsigned)inthostid;
1204 
1205 	return (0);
1206 }
1207 
1208 /*
1209  * sysctl helper function for kern.hostname and kern.domainnname.
1210  * resets the relevant recorded length when the underlying name is
1211  * changed.
1212  */
1213 static int
1214 sysctl_setlen(SYSCTLFN_ARGS)
1215 {
1216 	int error;
1217 
1218 	error = sysctl_lookup(SYSCTLFN_CALL(rnode));
1219 	if (error || newp == NULL)
1220 		return (error);
1221 
1222 	switch (rnode->sysctl_num) {
1223 	case KERN_HOSTNAME:
1224 		hostnamelen = strlen((const char*)rnode->sysctl_data);
1225 		break;
1226 	case KERN_DOMAINNAME:
1227 		domainnamelen = strlen((const char*)rnode->sysctl_data);
1228 		break;
1229 	}
1230 
1231 	return (0);
1232 }
1233 
1234 /*
1235  * sysctl helper routine for kern.clockrate. Assembles a struct on
1236  * the fly to be returned to the caller.
1237  */
1238 static int
1239 sysctl_kern_clockrate(SYSCTLFN_ARGS)
1240 {
1241 	struct clockinfo clkinfo;
1242 	struct sysctlnode node;
1243 
1244 	clkinfo.tick = tick;
1245 	clkinfo.tickadj = tickadj;
1246 	clkinfo.hz = hz;
1247 	clkinfo.profhz = profhz;
1248 	clkinfo.stathz = stathz ? stathz : hz;
1249 
1250 	node = *rnode;
1251 	node.sysctl_data = &clkinfo;
1252 	return (sysctl_lookup(SYSCTLFN_CALL(&node)));
1253 }
1254 
1255 
1256 /*
1257  * sysctl helper routine for kern.file pseudo-subtree.
1258  */
1259 static int
1260 sysctl_kern_file(SYSCTLFN_ARGS)
1261 {
1262 	int error;
1263 	size_t buflen;
1264 	struct file *fp, *dp, *np, fbuf;
1265 	char *start, *where;
1266 
1267 	start = where = oldp;
1268 	buflen = *oldlenp;
1269 	dp = NULL;
1270 
1271 	if (where == NULL) {
1272 		/*
1273 		 * overestimate by 10 files
1274 		 */
1275 		*oldlenp = sizeof(filehead) + (nfiles + 10) *
1276 		    sizeof(struct file);
1277 		return (0);
1278 	}
1279 
1280 	/*
1281 	 * first dcopyout filehead
1282 	 */
1283 	if (buflen < sizeof(filehead)) {
1284 		*oldlenp = 0;
1285 		return (0);
1286 	}
1287 	sysctl_unlock();
1288 	error = dcopyout(l, &filehead, where, sizeof(filehead));
1289 	if (error) {
1290 	 	sysctl_relock();
1291 		return error;
1292 	}
1293 	buflen -= sizeof(filehead);
1294 	where += sizeof(filehead);
1295 
1296 	/*
1297 	 * allocate dummy file descriptor to make position in list
1298 	 */
1299 	if ((dp = fgetdummy()) == NULL) {
1300 	 	sysctl_relock();
1301 		return ENOMEM;
1302 	}
1303 
1304 	/*
1305 	 * followed by an array of file structures
1306 	 */
1307 	mutex_enter(&filelist_lock);
1308 	for (fp = LIST_FIRST(&filehead); fp != NULL; fp = np) {
1309 	    	np = LIST_NEXT(fp, f_list);
1310 	    	mutex_enter(&fp->f_lock);
1311 	    	if (fp->f_count == 0) {
1312 		    	mutex_exit(&fp->f_lock);
1313 	    		continue;
1314 		}
1315 		/*
1316 		 * XXX Need to prevent that from being an alternative way
1317 		 * XXX to getting process information.
1318 		 */
1319 		if (kauth_authorize_generic(l->l_cred,
1320 		    KAUTH_GENERIC_CANSEE, fp->f_cred) != 0) {
1321 		    	mutex_exit(&fp->f_lock);
1322 			continue;
1323 		}
1324 		if (buflen < sizeof(struct file)) {
1325 			*oldlenp = where - start;
1326 		    	mutex_exit(&fp->f_lock);
1327 			error = ENOMEM;
1328 			break;
1329 		}
1330 		memcpy(&fbuf, fp, sizeof(fbuf));
1331 		LIST_INSERT_AFTER(fp, dp, f_list);
1332 	    	mutex_exit(&fp->f_lock);
1333 		mutex_exit(&filelist_lock);
1334 		error = dcopyout(l, &fbuf, where, sizeof(fbuf));
1335 		if (error) {
1336 			mutex_enter(&filelist_lock);
1337 			LIST_REMOVE(dp, f_list);
1338 			break;
1339 		}
1340 		buflen -= sizeof(struct file);
1341 		where += sizeof(struct file);
1342 		mutex_enter(&filelist_lock);
1343 		np = LIST_NEXT(dp, f_list);
1344 		LIST_REMOVE(dp, f_list);
1345 	}
1346 	mutex_exit(&filelist_lock);
1347 	*oldlenp = where - start;
1348  	if (dp != NULL)
1349 		fputdummy(dp);
1350  	sysctl_relock();
1351 	return (error);
1352 }
1353 
1354 /*
1355  * sysctl helper routine for kern.msgbufsize and kern.msgbuf. For the
1356  * former it merely checks the message buffer is set up. For the latter,
1357  * it also copies out the data if necessary.
1358  */
1359 static int
1360 sysctl_msgbuf(SYSCTLFN_ARGS)
1361 {
1362 	char *where = oldp;
1363 	size_t len, maxlen;
1364 	long beg, end;
1365 	extern kmutex_t log_lock;
1366 	int error;
1367 
1368 	if (!msgbufenabled || msgbufp->msg_magic != MSG_MAGIC) {
1369 		msgbufenabled = 0;
1370 		return (ENXIO);
1371 	}
1372 
1373 	switch (rnode->sysctl_num) {
1374 	case KERN_MSGBUFSIZE: {
1375 		struct sysctlnode node = *rnode;
1376 		int msg_bufs = (int)msgbufp->msg_bufs;
1377 		node.sysctl_data = &msg_bufs;
1378 		return (sysctl_lookup(SYSCTLFN_CALL(&node)));
1379 	}
1380 	case KERN_MSGBUF:
1381 		break;
1382 	default:
1383 		return (EOPNOTSUPP);
1384 	}
1385 
1386 	if (newp != NULL)
1387 		return (EPERM);
1388 
1389 	if (oldp == NULL) {
1390 		/* always return full buffer size */
1391 		*oldlenp = msgbufp->msg_bufs;
1392 		return (0);
1393 	}
1394 
1395 	sysctl_unlock();
1396 
1397 	/*
1398 	 * First, copy from the write pointer to the end of
1399 	 * message buffer.
1400 	 */
1401 	error = 0;
1402 	mutex_spin_enter(&log_lock);
1403 	maxlen = MIN(msgbufp->msg_bufs, *oldlenp);
1404 	beg = msgbufp->msg_bufx;
1405 	end = msgbufp->msg_bufs;
1406 	mutex_spin_exit(&log_lock);
1407 
1408 	while (maxlen > 0) {
1409 		len = MIN(end - beg, maxlen);
1410 		if (len == 0)
1411 			break;
1412 		/* XXX unlocked, but hardly matters. */
1413 		error = dcopyout(l, &msgbufp->msg_bufc[beg], where, len);
1414 		if (error)
1415 			break;
1416 		where += len;
1417 		maxlen -= len;
1418 
1419 		/*
1420 		 * ... then, copy from the beginning of message buffer to
1421 		 * the write pointer.
1422 		 */
1423 		beg = 0;
1424 		end = msgbufp->msg_bufx;
1425 	}
1426 
1427 	sysctl_relock();
1428 	return (error);
1429 }
1430 
1431 /*
1432  * sysctl helper routine for kern.defcorename. In the case of a new
1433  * string being assigned, check that it's not a zero-length string.
1434  * (XXX the check in -current doesn't work, but do we really care?)
1435  */
1436 static int
1437 sysctl_kern_defcorename(SYSCTLFN_ARGS)
1438 {
1439 	int error;
1440 	char *newcorename;
1441 	struct sysctlnode node;
1442 
1443 	newcorename = PNBUF_GET();
1444 	node = *rnode;
1445 	node.sysctl_data = &newcorename[0];
1446 	memcpy(node.sysctl_data, rnode->sysctl_data, MAXPATHLEN);
1447 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1448 	if (error || newp == NULL) {
1449 		goto done;
1450 	}
1451 
1452 	/*
1453 	 * when sysctl_lookup() deals with a string, it's guaranteed
1454 	 * to come back nul terminated. So there.  :)
1455 	 */
1456 	if (strlen(newcorename) == 0) {
1457 		error = EINVAL;
1458 	} else {
1459 		memcpy(rnode->sysctl_data, node.sysctl_data, MAXPATHLEN);
1460 		error = 0;
1461 	}
1462 done:
1463 	PNBUF_PUT(newcorename);
1464 	return error;
1465 }
1466 
1467 /*
1468  * sysctl helper routine for kern.cp_time node. Adds up cpu time
1469  * across all cpus.
1470  */
1471 static int
1472 sysctl_kern_cptime(SYSCTLFN_ARGS)
1473 {
1474 	struct sysctlnode node = *rnode;
1475 	uint64_t *cp_time = NULL;
1476 	int error, n = ncpu, i;
1477 	struct cpu_info *ci;
1478 	CPU_INFO_ITERATOR cii;
1479 
1480 	/*
1481 	 * if you specifically pass a buffer that is the size of the
1482 	 * sum, or if you are probing for the size, you get the "sum"
1483 	 * of cp_time (and the size thereof) across all processors.
1484 	 *
1485 	 * alternately, you can pass an additional mib number and get
1486 	 * cp_time for that particular processor.
1487 	 */
1488 	switch (namelen) {
1489 	case 0:
1490 		if (*oldlenp == sizeof(uint64_t) * CPUSTATES || oldp == NULL) {
1491 			node.sysctl_size = sizeof(uint64_t) * CPUSTATES;
1492 			n = -1; /* SUM */
1493 		}
1494 		else {
1495 			node.sysctl_size = n * sizeof(uint64_t) * CPUSTATES;
1496 			n = -2; /* ALL */
1497 		}
1498 		break;
1499 	case 1:
1500 		if (name[0] < 0 || name[0] >= n)
1501 			return (ENOENT); /* ENOSUCHPROCESSOR */
1502 		node.sysctl_size = sizeof(uint64_t) * CPUSTATES;
1503 		n = name[0];
1504 		/*
1505 		 * adjust these so that sysctl_lookup() will be happy
1506 		 */
1507 		name++;
1508 		namelen--;
1509 		break;
1510 	default:
1511 		return (EINVAL);
1512 	}
1513 
1514 	cp_time = kmem_alloc(node.sysctl_size, KM_SLEEP);
1515 	if (cp_time == NULL)
1516 		return (ENOMEM);
1517 	node.sysctl_data = cp_time;
1518 	memset(cp_time, 0, node.sysctl_size);
1519 
1520 	for (CPU_INFO_FOREACH(cii, ci)) {
1521 		if (n <= 0) {
1522 			for (i = 0; i < CPUSTATES; i++) {
1523 				cp_time[i] += ci->ci_schedstate.spc_cp_time[i];
1524 			}
1525 		}
1526 		/*
1527 		 * if a specific processor was requested and we just
1528 		 * did it, we're done here
1529 		 */
1530 		if (n == 0)
1531 			break;
1532 		/*
1533 		 * if doing "all", skip to next cp_time set for next processor
1534 		 */
1535 		if (n == -2)
1536 			cp_time += CPUSTATES;
1537 		/*
1538 		 * if we're doing a specific processor, we're one
1539 		 * processor closer
1540 		 */
1541 		if (n > 0)
1542 			n--;
1543 	}
1544 
1545 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1546 	kmem_free(node.sysctl_data, node.sysctl_size);
1547 	return (error);
1548 }
1549 
1550 #if NPTY > 0
1551 /*
1552  * sysctl helper routine for kern.maxptys. Ensures that any new value
1553  * is acceptable to the pty subsystem.
1554  */
1555 static int
1556 sysctl_kern_maxptys(SYSCTLFN_ARGS)
1557 {
1558 	int pty_maxptys(int, int);		/* defined in kern/tty_pty.c */
1559 	int error, xmax;
1560 	struct sysctlnode node;
1561 
1562 	/* get current value of maxptys */
1563 	xmax = pty_maxptys(0, 0);
1564 
1565 	node = *rnode;
1566 	node.sysctl_data = &xmax;
1567 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1568 	if (error || newp == NULL)
1569 		return (error);
1570 
1571 	if (xmax != pty_maxptys(xmax, 1))
1572 		return (EINVAL);
1573 
1574 	return (0);
1575 }
1576 #endif /* NPTY > 0 */
1577 
1578 /*
1579  * sysctl helper routine for kern.sbmax. Basically just ensures that
1580  * any new value is not too small.
1581  */
1582 static int
1583 sysctl_kern_sbmax(SYSCTLFN_ARGS)
1584 {
1585 	int error, new_sbmax;
1586 	struct sysctlnode node;
1587 
1588 	new_sbmax = sb_max;
1589 	node = *rnode;
1590 	node.sysctl_data = &new_sbmax;
1591 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1592 	if (error || newp == NULL)
1593 		return (error);
1594 
1595 	KERNEL_LOCK(1, NULL);
1596 	error = sb_max_set(new_sbmax);
1597 	KERNEL_UNLOCK_ONE(NULL);
1598 
1599 	return (error);
1600 }
1601 
1602 /*
1603  * sysctl helper routine for kern.urandom node. Picks a random number
1604  * for you.
1605  */
1606 static int
1607 sysctl_kern_urnd(SYSCTLFN_ARGS)
1608 {
1609 #if NRND > 0
1610 	int v, rv;
1611 
1612 	KERNEL_LOCK(1, NULL);
1613 	rv = rnd_extract_data(&v, sizeof(v), RND_EXTRACT_ANY);
1614 	KERNEL_UNLOCK_ONE(NULL);
1615 	if (rv == sizeof(v)) {
1616 		struct sysctlnode node = *rnode;
1617 		node.sysctl_data = &v;
1618 		return (sysctl_lookup(SYSCTLFN_CALL(&node)));
1619 	}
1620 	else
1621 		return (EIO);	/*XXX*/
1622 #else
1623 	return (EOPNOTSUPP);
1624 #endif
1625 }
1626 
1627 /*
1628  * sysctl helper routine for kern.arandom node. Picks a random number
1629  * for you.
1630  */
1631 static int
1632 sysctl_kern_arnd(SYSCTLFN_ARGS)
1633 {
1634 #if NRND > 0
1635 	int error;
1636 	void *v;
1637 	struct sysctlnode node = *rnode;
1638 
1639 	if (*oldlenp == 0)
1640 		return 0;
1641 	if (*oldlenp > 8192)
1642 		return E2BIG;
1643 
1644 	v = kmem_alloc(*oldlenp, KM_SLEEP);
1645 	arc4randbytes(v, *oldlenp);
1646 	node.sysctl_data = v;
1647 	node.sysctl_size = *oldlenp;
1648 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1649 	kmem_free(v, *oldlenp);
1650 	return error;
1651 #else
1652 	return (EOPNOTSUPP);
1653 #endif
1654 }
1655 /*
1656  * sysctl helper routine to do kern.lwp.* work.
1657  */
1658 static int
1659 sysctl_kern_lwp(SYSCTLFN_ARGS)
1660 {
1661 	struct kinfo_lwp klwp;
1662 	struct proc *p;
1663 	struct lwp *l2, *l3;
1664 	char *where, *dp;
1665 	int pid, elem_size, elem_count;
1666 	int buflen, needed, error;
1667 	bool gotit;
1668 
1669 	if (namelen == 1 && name[0] == CTL_QUERY)
1670 		return (sysctl_query(SYSCTLFN_CALL(rnode)));
1671 
1672 	dp = where = oldp;
1673 	buflen = where != NULL ? *oldlenp : 0;
1674 	error = needed = 0;
1675 
1676 	if (newp != NULL || namelen != 3)
1677 		return (EINVAL);
1678 	pid = name[0];
1679 	elem_size = name[1];
1680 	elem_count = name[2];
1681 
1682 	sysctl_unlock();
1683 	if (pid == -1) {
1684 		mutex_enter(proc_lock);
1685 		LIST_FOREACH(p, &allproc, p_list) {
1686 			/* Grab a hold on the process. */
1687 			if (!rw_tryenter(&p->p_reflock, RW_READER)) {
1688 				continue;
1689 			}
1690 			mutex_exit(proc_lock);
1691 
1692 			mutex_enter(p->p_lock);
1693 			LIST_FOREACH(l2, &p->p_lwps, l_sibling) {
1694 				if (buflen >= elem_size && elem_count > 0) {
1695 					lwp_lock(l2);
1696 					fill_lwp(l2, &klwp);
1697 					lwp_unlock(l2);
1698 					mutex_exit(p->p_lock);
1699 
1700 					/*
1701 					 * Copy out elem_size, but not
1702 					 * larger than the size of a
1703 					 * struct kinfo_proc2.
1704 					 */
1705 					error = dcopyout(l, &klwp, dp,
1706 					    min(sizeof(klwp), elem_size));
1707 					if (error) {
1708 						rw_exit(&p->p_reflock);
1709 						goto cleanup;
1710 					}
1711 					mutex_enter(p->p_lock);
1712 					LIST_FOREACH(l3, &p->p_lwps,
1713 					    l_sibling) {
1714 						if (l2 == l3)
1715 							break;
1716 					}
1717 					if (l3 == NULL) {
1718 						mutex_exit(p->p_lock);
1719 						rw_exit(&p->p_reflock);
1720 						error = EAGAIN;
1721 						goto cleanup;
1722 					}
1723 					dp += elem_size;
1724 					buflen -= elem_size;
1725 					elem_count--;
1726 				}
1727 				needed += elem_size;
1728 			}
1729 			mutex_exit(p->p_lock);
1730 
1731 			/* Drop reference to process. */
1732 			mutex_enter(proc_lock);
1733 			rw_exit(&p->p_reflock);
1734 		}
1735 		mutex_exit(proc_lock);
1736 	} else {
1737 		mutex_enter(proc_lock);
1738 		p = p_find(pid, PFIND_LOCKED);
1739 		if (p == NULL) {
1740 			error = ESRCH;
1741 			mutex_exit(proc_lock);
1742 			goto cleanup;
1743 		}
1744 		/* Grab a hold on the process. */
1745 		gotit = rw_tryenter(&p->p_reflock, RW_READER);
1746 		mutex_exit(proc_lock);
1747 		if (!gotit) {
1748 			error = ESRCH;
1749 			goto cleanup;
1750 		}
1751 
1752 		mutex_enter(p->p_lock);
1753 		LIST_FOREACH(l2, &p->p_lwps, l_sibling) {
1754 			if (buflen >= elem_size && elem_count > 0) {
1755 				lwp_lock(l2);
1756 				fill_lwp(l2, &klwp);
1757 				lwp_unlock(l2);
1758 				mutex_exit(p->p_lock);
1759 				/*
1760 				 * Copy out elem_size, but not larger than
1761 				 * the size of a struct kinfo_proc2.
1762 				 */
1763 				error = dcopyout(l, &klwp, dp,
1764 				    min(sizeof(klwp), elem_size));
1765 				if (error) {
1766 					rw_exit(&p->p_reflock);
1767 					goto cleanup;
1768 				}
1769 				mutex_enter(p->p_lock);
1770 				LIST_FOREACH(l3, &p->p_lwps, l_sibling) {
1771 					if (l2 == l3)
1772 						break;
1773 				}
1774 				if (l3 == NULL) {
1775 					mutex_exit(p->p_lock);
1776 					rw_exit(&p->p_reflock);
1777 					error = EAGAIN;
1778 					goto cleanup;
1779 				}
1780 				dp += elem_size;
1781 				buflen -= elem_size;
1782 				elem_count--;
1783 			}
1784 			needed += elem_size;
1785 		}
1786 		mutex_exit(p->p_lock);
1787 
1788 		/* Drop reference to process. */
1789 		rw_exit(&p->p_reflock);
1790 	}
1791 
1792 	if (where != NULL) {
1793 		*oldlenp = dp - where;
1794 		if (needed > *oldlenp) {
1795 			sysctl_relock();
1796 			return (ENOMEM);
1797 		}
1798 	} else {
1799 		needed += KERN_LWPSLOP;
1800 		*oldlenp = needed;
1801 	}
1802 	error = 0;
1803  cleanup:
1804 	sysctl_relock();
1805 	return (error);
1806 }
1807 
1808 /*
1809  * sysctl helper routine for kern.forkfsleep node. Ensures that the
1810  * given value is not too large or two small, and is at least one
1811  * timer tick if not zero.
1812  */
1813 static int
1814 sysctl_kern_forkfsleep(SYSCTLFN_ARGS)
1815 {
1816 	/* userland sees value in ms, internally is in ticks */
1817 	extern int forkfsleep;		/* defined in kern/kern_fork.c */
1818 	int error, timo, lsleep;
1819 	struct sysctlnode node;
1820 
1821 	lsleep = forkfsleep * 1000 / hz;
1822 	node = *rnode;
1823 	node.sysctl_data = &lsleep;
1824 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1825 	if (error || newp == NULL)
1826 		return (error);
1827 
1828 	/* refuse negative values, and overly 'long time' */
1829 	if (lsleep < 0 || lsleep > MAXSLP * 1000)
1830 		return (EINVAL);
1831 
1832 	timo = mstohz(lsleep);
1833 
1834 	/* if the interval is >0 ms && <1 tick, use 1 tick */
1835 	if (lsleep != 0 && timo == 0)
1836 		forkfsleep = 1;
1837 	else
1838 		forkfsleep = timo;
1839 
1840 	return (0);
1841 }
1842 
1843 /*
1844  * sysctl helper routine for kern.root_partition
1845  */
1846 static int
1847 sysctl_kern_root_partition(SYSCTLFN_ARGS)
1848 {
1849 	int rootpart = DISKPART(rootdev);
1850 	struct sysctlnode node = *rnode;
1851 
1852 	node.sysctl_data = &rootpart;
1853 	return (sysctl_lookup(SYSCTLFN_CALL(&node)));
1854 }
1855 
1856 /*
1857  * sysctl helper function for kern.drivers
1858  */
1859 static int
1860 sysctl_kern_drivers(SYSCTLFN_ARGS)
1861 {
1862 	int error;
1863 	size_t buflen;
1864 	struct kinfo_drivers kd;
1865 	char *start, *where;
1866 	const char *dname;
1867 	int i;
1868 	extern struct devsw_conv *devsw_conv;
1869 	extern int max_devsw_convs;
1870 
1871 	if (newp != NULL || namelen != 0)
1872 		return (EINVAL);
1873 
1874 	start = where = oldp;
1875 	buflen = *oldlenp;
1876 	if (where == NULL) {
1877 		*oldlenp = max_devsw_convs * sizeof kd;
1878 		return 0;
1879 	}
1880 
1881 	/*
1882 	 * An array of kinfo_drivers structures
1883 	 */
1884 	error = 0;
1885 	sysctl_unlock();
1886 	mutex_enter(&specfs_lock);
1887 	for (i = 0; i < max_devsw_convs; i++) {
1888 		dname = devsw_conv[i].d_name;
1889 		if (dname == NULL)
1890 			continue;
1891 		if (buflen < sizeof kd) {
1892 			error = ENOMEM;
1893 			break;
1894 		}
1895 		memset(&kd, 0, sizeof(kd));
1896 		kd.d_bmajor = devsw_conv[i].d_bmajor;
1897 		kd.d_cmajor = devsw_conv[i].d_cmajor;
1898 		strlcpy(kd.d_name, dname, sizeof kd.d_name);
1899 		mutex_exit(&specfs_lock);
1900 		error = dcopyout(l, &kd, where, sizeof kd);
1901 		mutex_enter(&specfs_lock);
1902 		if (error != 0)
1903 			break;
1904 		buflen -= sizeof kd;
1905 		where += sizeof kd;
1906 	}
1907 	mutex_exit(&specfs_lock);
1908 	sysctl_relock();
1909 	*oldlenp = where - start;
1910 	return error;
1911 }
1912 
1913 /*
1914  * sysctl helper function for kern.file2
1915  */
1916 static int
1917 sysctl_kern_file2(SYSCTLFN_ARGS)
1918 {
1919 	struct proc *p;
1920 	struct file *fp, *tp, *np;
1921 	struct filedesc *fd;
1922 	struct kinfo_file kf;
1923 	char *dp;
1924 	u_int i, op;
1925 	size_t len, needed, elem_size, out_size;
1926 	int error, arg, elem_count;
1927 	fdfile_t *ff;
1928 
1929 	if (namelen == 1 && name[0] == CTL_QUERY)
1930 		return (sysctl_query(SYSCTLFN_CALL(rnode)));
1931 
1932 	if (namelen != 4)
1933 		return (EINVAL);
1934 
1935 	error = 0;
1936 	dp = oldp;
1937 	len = (oldp != NULL) ? *oldlenp : 0;
1938 	op = name[0];
1939 	arg = name[1];
1940 	elem_size = name[2];
1941 	elem_count = name[3];
1942 	out_size = MIN(sizeof(kf), elem_size);
1943 	needed = 0;
1944 
1945 	if (elem_size < 1 || elem_count < 0)
1946 		return (EINVAL);
1947 
1948 	switch (op) {
1949 	case KERN_FILE_BYFILE:
1950 		/*
1951 		 * doesn't use arg so it must be zero
1952 		 */
1953 		if (arg != 0)
1954 			return (EINVAL);
1955 		sysctl_unlock();
1956 		/*
1957 		 * allocate dummy file descriptor to make position in list
1958 		 */
1959 		if ((tp = fgetdummy()) == NULL) {
1960 		 	sysctl_relock();
1961 			return ENOMEM;
1962 		}
1963 		mutex_enter(&filelist_lock);
1964 		for (fp = LIST_FIRST(&filehead); fp != NULL; fp = np) {
1965 			np = LIST_NEXT(fp, f_list);
1966 			mutex_enter(&fp->f_lock);
1967 			if (fp->f_count == 0) {
1968 				mutex_exit(&fp->f_lock);
1969 				continue;
1970 			}
1971 			/*
1972 			 * XXX Need to prevent that from being an alternative
1973 			 * XXX way for getting process information.
1974 			 */
1975 			if (kauth_authorize_generic(l->l_cred,
1976 			    KAUTH_GENERIC_CANSEE, fp->f_cred) != 0) {
1977 				mutex_exit(&fp->f_lock);
1978 				continue;
1979 			}
1980 			if (len >= elem_size && elem_count > 0) {
1981 				fill_file(&kf, fp, NULL, 0, 0);
1982 				LIST_INSERT_AFTER(fp, tp, f_list);
1983 				mutex_exit(&fp->f_lock);
1984 				mutex_exit(&filelist_lock);
1985 				error = dcopyout(l, &kf, dp, out_size);
1986 				mutex_enter(&filelist_lock);
1987 				np = LIST_NEXT(tp, f_list);
1988 				LIST_REMOVE(tp, f_list);
1989 				if (error) {
1990 					break;
1991 				}
1992 				dp += elem_size;
1993 				len -= elem_size;
1994 			} else {
1995 				mutex_exit(&fp->f_lock);
1996 			}
1997 			if (elem_count > 0) {
1998 				needed += elem_size;
1999 				if (elem_count != INT_MAX)
2000 					elem_count--;
2001 			}
2002 		}
2003 		mutex_exit(&filelist_lock);
2004 		fputdummy(tp);
2005 		sysctl_relock();
2006 		break;
2007 	case KERN_FILE_BYPID:
2008 		if (arg < -1)
2009 			/* -1 means all processes */
2010 			return (EINVAL);
2011 		sysctl_unlock();
2012 		mutex_enter(proc_lock);
2013 		LIST_FOREACH(p, &allproc, p_list) {
2014 			if (p->p_stat == SIDL) {
2015 				/* skip embryonic processes */
2016 				continue;
2017 			}
2018 			if (arg > 0 && p->p_pid != arg) {
2019 				/* pick only the one we want */
2020 				/* XXX want 0 to mean "kernel files" */
2021 				continue;
2022 			}
2023 			mutex_enter(p->p_lock);
2024 			error = kauth_authorize_process(l->l_cred,
2025 			    KAUTH_PROCESS_CANSEE, p,
2026 			    KAUTH_ARG(KAUTH_REQ_PROCESS_CANSEE_OPENFILES),
2027 			    NULL, NULL);
2028 			mutex_exit(p->p_lock);
2029 			if (error != 0) {
2030 				continue;
2031 			}
2032 
2033 			/*
2034 			 * Grab a hold on the process.
2035 			 */
2036 			if (!rw_tryenter(&p->p_reflock, RW_READER)) {
2037 				continue;
2038 			}
2039 			mutex_exit(proc_lock);
2040 
2041 			/* XXX Do we need to check permission per file? */
2042 			fd = p->p_fd;
2043 			mutex_enter(&fd->fd_lock);
2044 			for (i = 0; i < fd->fd_nfiles; i++) {
2045 				if ((ff = fd->fd_ofiles[i]) == NULL) {
2046 					continue;
2047 				}
2048 				mutex_enter(&ff->ff_lock);
2049 				if ((fp = ff->ff_file) == NULL) {
2050 					mutex_exit(&ff->ff_lock);
2051 					continue;
2052 				}
2053 				if (len >= elem_size && elem_count > 0) {
2054 					mutex_enter(&fp->f_lock);
2055 					fill_file(&kf, fp, ff, i, p->p_pid);
2056 					mutex_exit(&fp->f_lock);
2057 					mutex_exit(&ff->ff_lock);
2058 					mutex_exit(&fd->fd_lock);
2059 					error = dcopyout(l, &kf, dp, out_size);
2060 					mutex_enter(&fd->fd_lock);
2061 					if (error)
2062 						break;
2063 					dp += elem_size;
2064 					len -= elem_size;
2065 				} else {
2066 					mutex_exit(&ff->ff_lock);
2067 				}
2068 				if (elem_count > 0) {
2069 					needed += elem_size;
2070 					if (elem_count != INT_MAX)
2071 						elem_count--;
2072 				}
2073 			}
2074 			mutex_exit(&fd->fd_lock);
2075 
2076 			/*
2077 			 * Release reference to process.
2078 			 */
2079 			mutex_enter(proc_lock);
2080 			rw_exit(&p->p_reflock);
2081 		}
2082 		mutex_exit(proc_lock);
2083 		sysctl_relock();
2084 		break;
2085 	default:
2086 		return (EINVAL);
2087 	}
2088 
2089 	if (oldp == NULL)
2090 		needed += KERN_FILESLOP * elem_size;
2091 	*oldlenp = needed;
2092 
2093 	return (error);
2094 }
2095 
2096 static void
2097 fill_file(struct kinfo_file *kp, const file_t *fp, const fdfile_t *ff,
2098 	  int i, pid_t pid)
2099 {
2100 
2101 	memset(kp, 0, sizeof(*kp));
2102 
2103 	kp->ki_fileaddr =	PTRTOUINT64(fp);
2104 	kp->ki_flag =		fp->f_flag;
2105 	kp->ki_iflags =		fp->f_iflags;
2106 	kp->ki_ftype =		fp->f_type;
2107 	kp->ki_count =		fp->f_count;
2108 	kp->ki_msgcount =	fp->f_msgcount;
2109 	kp->ki_fucred =		PTRTOUINT64(fp->f_cred);
2110 	kp->ki_fuid =		kauth_cred_geteuid(fp->f_cred);
2111 	kp->ki_fgid =		kauth_cred_getegid(fp->f_cred);
2112 	kp->ki_fops =		PTRTOUINT64(fp->f_ops);
2113 	kp->ki_foffset =	fp->f_offset;
2114 	kp->ki_fdata =		PTRTOUINT64(fp->f_data);
2115 
2116 	/* vnode information to glue this file to something */
2117 	if (fp->f_type == DTYPE_VNODE) {
2118 		struct vnode *vp = (struct vnode *)fp->f_data;
2119 
2120 		kp->ki_vun =	PTRTOUINT64(vp->v_un.vu_socket);
2121 		kp->ki_vsize =	vp->v_size;
2122 		kp->ki_vtype =	vp->v_type;
2123 		kp->ki_vtag =	vp->v_tag;
2124 		kp->ki_vdata =	PTRTOUINT64(vp->v_data);
2125 	}
2126 
2127 	/* process information when retrieved via KERN_FILE_BYPID */
2128 	if (ff != NULL) {
2129 		kp->ki_pid =		pid;
2130 		kp->ki_fd =		i;
2131 		kp->ki_ofileflags =	ff->ff_exclose;
2132 		kp->ki_usecount =	ff->ff_refcnt;
2133 	}
2134 }
2135 
2136 static int
2137 sysctl_doeproc(SYSCTLFN_ARGS)
2138 {
2139 	struct eproc *eproc;
2140 	struct kinfo_proc2 *kproc2;
2141 	struct kinfo_proc *dp;
2142 	struct proc *p, *next, *marker;
2143 	char *where, *dp2;
2144 	int type, op, arg, error;
2145 	u_int elem_size, elem_count;
2146 	size_t buflen, needed;
2147 	bool match, zombie, mmmbrains;
2148 
2149 	if (namelen == 1 && name[0] == CTL_QUERY)
2150 		return (sysctl_query(SYSCTLFN_CALL(rnode)));
2151 
2152 	dp = oldp;
2153 	dp2 = where = oldp;
2154 	buflen = where != NULL ? *oldlenp : 0;
2155 	error = 0;
2156 	needed = 0;
2157 	type = rnode->sysctl_num;
2158 
2159 	if (type == KERN_PROC) {
2160 		if (namelen != 2 && !(namelen == 1 && name[0] == KERN_PROC_ALL))
2161 			return (EINVAL);
2162 		op = name[0];
2163 		if (op != KERN_PROC_ALL)
2164 			arg = name[1];
2165 		else
2166 			arg = 0;		/* Quell compiler warning */
2167 		elem_size = elem_count = 0;	/* Ditto */
2168 	} else {
2169 		if (namelen != 4)
2170 			return (EINVAL);
2171 		op = name[0];
2172 		arg = name[1];
2173 		elem_size = name[2];
2174 		elem_count = name[3];
2175 	}
2176 
2177 	sysctl_unlock();
2178 
2179 	if (type == KERN_PROC) {
2180 		eproc = kmem_alloc(sizeof(*eproc), KM_SLEEP);
2181 		kproc2 = NULL;
2182 	} else {
2183 		eproc = NULL;
2184 		kproc2 = kmem_alloc(sizeof(*kproc2), KM_SLEEP);
2185 	}
2186 	marker = kmem_alloc(sizeof(*marker), KM_SLEEP);
2187 
2188 	mutex_enter(proc_lock);
2189 	mmmbrains = false;
2190 	for (p = LIST_FIRST(&allproc);; p = next) {
2191 		if (p == NULL) {
2192 			if (!mmmbrains) {
2193 				p = LIST_FIRST(&zombproc);
2194 				mmmbrains = true;
2195 			}
2196 			if (p == NULL)
2197 				break;
2198 		}
2199 		next = LIST_NEXT(p, p_list);
2200 
2201 		/*
2202 		 * Skip embryonic processes.
2203 		 */
2204 		if (p->p_stat == SIDL)
2205 			continue;
2206 
2207 		mutex_enter(p->p_lock);
2208 		error = kauth_authorize_process(l->l_cred,
2209 		    KAUTH_PROCESS_CANSEE, p,
2210 		    KAUTH_ARG(KAUTH_REQ_PROCESS_CANSEE_ENTRY), NULL, NULL);
2211 		if (error != 0) {
2212 			mutex_exit(p->p_lock);
2213 			continue;
2214 		}
2215 
2216 		/*
2217 		 * TODO - make more efficient (see notes below).
2218 		 * do by session.
2219 		 */
2220 		switch (op) {
2221 		case KERN_PROC_PID:
2222 			/* could do this with just a lookup */
2223 			match = (p->p_pid == (pid_t)arg);
2224 			break;
2225 
2226 		case KERN_PROC_PGRP:
2227 			/* could do this by traversing pgrp */
2228 			match = (p->p_pgrp->pg_id == (pid_t)arg);
2229 			break;
2230 
2231 		case KERN_PROC_SESSION:
2232 			match = (p->p_session->s_sid == (pid_t)arg);
2233 			break;
2234 
2235 		case KERN_PROC_TTY:
2236 			match = true;
2237 			if (arg == (int) KERN_PROC_TTY_REVOKE) {
2238 				if ((p->p_lflag & PL_CONTROLT) == 0 ||
2239 				    p->p_session->s_ttyp == NULL ||
2240 				    p->p_session->s_ttyvp != NULL) {
2241 				    	match = false;
2242 				}
2243 			} else if ((p->p_lflag & PL_CONTROLT) == 0 ||
2244 			    p->p_session->s_ttyp == NULL) {
2245 				if ((dev_t)arg != KERN_PROC_TTY_NODEV) {
2246 					match = false;
2247 				}
2248 			} else if (p->p_session->s_ttyp->t_dev != (dev_t)arg) {
2249 				match = false;
2250 			}
2251 			break;
2252 
2253 		case KERN_PROC_UID:
2254 			match = (kauth_cred_geteuid(p->p_cred) == (uid_t)arg);
2255 			break;
2256 
2257 		case KERN_PROC_RUID:
2258 			match = (kauth_cred_getuid(p->p_cred) == (uid_t)arg);
2259 			break;
2260 
2261 		case KERN_PROC_GID:
2262 			match = (kauth_cred_getegid(p->p_cred) == (uid_t)arg);
2263 			break;
2264 
2265 		case KERN_PROC_RGID:
2266 			match = (kauth_cred_getgid(p->p_cred) == (uid_t)arg);
2267 			break;
2268 
2269 		case KERN_PROC_ALL:
2270 			match = true;
2271 			/* allow everything */
2272 			break;
2273 
2274 		default:
2275 			error = EINVAL;
2276 			mutex_exit(p->p_lock);
2277 			goto cleanup;
2278 		}
2279 		if (!match) {
2280 			mutex_exit(p->p_lock);
2281 			continue;
2282 		}
2283 
2284 		/*
2285 		 * Grab a hold on the process.
2286 		 */
2287 		if (mmmbrains) {
2288 			zombie = true;
2289 		} else {
2290 			zombie = !rw_tryenter(&p->p_reflock, RW_READER);
2291 		}
2292 		if (zombie) {
2293 			LIST_INSERT_AFTER(p, marker, p_list);
2294 		}
2295 
2296 		if (type == KERN_PROC) {
2297 			if (buflen >= sizeof(struct kinfo_proc)) {
2298 				fill_eproc(p, eproc, zombie);
2299 				mutex_exit(p->p_lock);
2300 				mutex_exit(proc_lock);
2301 				error = dcopyout(l, p, &dp->kp_proc,
2302 				    sizeof(struct proc));
2303 				mutex_enter(proc_lock);
2304 				if (error) {
2305 					goto bah;
2306 				}
2307 				error = dcopyout(l, eproc, &dp->kp_eproc,
2308 				    sizeof(*eproc));
2309 				if (error) {
2310 					goto bah;
2311 				}
2312 				dp++;
2313 				buflen -= sizeof(struct kinfo_proc);
2314 			} else {
2315 				mutex_exit(p->p_lock);
2316 			}
2317 			needed += sizeof(struct kinfo_proc);
2318 		} else { /* KERN_PROC2 */
2319 			if (buflen >= elem_size && elem_count > 0) {
2320 				fill_kproc2(p, kproc2, zombie);
2321 				mutex_exit(p->p_lock);
2322 				mutex_exit(proc_lock);
2323 				/*
2324 				 * Copy out elem_size, but not larger than
2325 				 * the size of a struct kinfo_proc2.
2326 				 */
2327 				error = dcopyout(l, kproc2, dp2,
2328 				    min(sizeof(*kproc2), elem_size));
2329 				mutex_enter(proc_lock);
2330 				if (error) {
2331 					goto bah;
2332 				}
2333 				dp2 += elem_size;
2334 				buflen -= elem_size;
2335 				elem_count--;
2336 			} else {
2337 				mutex_exit(p->p_lock);
2338 			}
2339 			needed += elem_size;
2340 		}
2341 
2342 		/*
2343 		 * Release reference to process.
2344 		 */
2345 	 	if (zombie) {
2346 			next = LIST_NEXT(marker, p_list);
2347  			LIST_REMOVE(marker, p_list);
2348 		} else {
2349 			rw_exit(&p->p_reflock);
2350 		}
2351 	}
2352 	mutex_exit(proc_lock);
2353 
2354 	if (where != NULL) {
2355 		if (type == KERN_PROC)
2356 			*oldlenp = (char *)dp - where;
2357 		else
2358 			*oldlenp = dp2 - where;
2359 		if (needed > *oldlenp) {
2360 			error = ENOMEM;
2361 			goto out;
2362 		}
2363 	} else {
2364 		needed += KERN_PROCSLOP;
2365 		*oldlenp = needed;
2366 	}
2367 	if (kproc2)
2368 		kmem_free(kproc2, sizeof(*kproc2));
2369 	if (eproc)
2370 		kmem_free(eproc, sizeof(*eproc));
2371 	if (marker)
2372 		kmem_free(marker, sizeof(*marker));
2373 	sysctl_relock();
2374 	return 0;
2375  bah:
2376  	if (zombie)
2377  		LIST_REMOVE(marker, p_list);
2378 	else
2379 		rw_exit(&p->p_reflock);
2380  cleanup:
2381 	mutex_exit(proc_lock);
2382  out:
2383 	if (kproc2)
2384 		kmem_free(kproc2, sizeof(*kproc2));
2385 	if (eproc)
2386 		kmem_free(eproc, sizeof(*eproc));
2387 	if (marker)
2388 		kmem_free(marker, sizeof(*marker));
2389 	sysctl_relock();
2390 	return error;
2391 }
2392 
2393 /*
2394  * sysctl helper routine for kern.proc_args pseudo-subtree.
2395  */
2396 static int
2397 sysctl_kern_proc_args(SYSCTLFN_ARGS)
2398 {
2399 	struct ps_strings pss;
2400 	struct proc *p;
2401 	size_t len, i;
2402 	struct uio auio;
2403 	struct iovec aiov;
2404 	pid_t pid;
2405 	int nargv, type, error, argvlen;
2406 	char *arg;
2407 	char **argv = NULL;
2408 	char *tmp;
2409 	struct vmspace *vmspace;
2410 	vaddr_t psstr_addr;
2411 	vaddr_t offsetn;
2412 	vaddr_t offsetv;
2413 
2414 	if (namelen == 1 && name[0] == CTL_QUERY)
2415 		return (sysctl_query(SYSCTLFN_CALL(rnode)));
2416 
2417 	if (newp != NULL || namelen != 2)
2418 		return (EINVAL);
2419 	pid = name[0];
2420 	type = name[1];
2421 	argv = NULL;
2422 	argvlen = 0;
2423 
2424 	switch (type) {
2425 	case KERN_PROC_ARGV:
2426 	case KERN_PROC_NARGV:
2427 	case KERN_PROC_ENV:
2428 	case KERN_PROC_NENV:
2429 		/* ok */
2430 		break;
2431 	default:
2432 		return (EINVAL);
2433 	}
2434 
2435 	sysctl_unlock();
2436 
2437 	/* check pid */
2438 	mutex_enter(proc_lock);
2439 	if ((p = p_find(pid, PFIND_LOCKED)) == NULL) {
2440 		error = EINVAL;
2441 		goto out_locked;
2442 	}
2443 	mutex_enter(p->p_lock);
2444 
2445 	/* Check permission. */
2446 	if (type == KERN_PROC_ARGV || type == KERN_PROC_NARGV)
2447 		error = kauth_authorize_process(l->l_cred, KAUTH_PROCESS_CANSEE,
2448 		    p, KAUTH_ARG(KAUTH_REQ_PROCESS_CANSEE_ARGS), NULL, NULL);
2449 	else if (type == KERN_PROC_ENV || type == KERN_PROC_NENV)
2450 		error = kauth_authorize_process(l->l_cred, KAUTH_PROCESS_CANSEE,
2451 		    p, KAUTH_ARG(KAUTH_REQ_PROCESS_CANSEE_ENV), NULL, NULL);
2452 	else
2453 		error = EINVAL; /* XXXGCC */
2454 	if (error) {
2455 		mutex_exit(p->p_lock);
2456 		goto out_locked;
2457 	}
2458 
2459 	if (oldp == NULL) {
2460 		if (type == KERN_PROC_NARGV || type == KERN_PROC_NENV)
2461 			*oldlenp = sizeof (int);
2462 		else
2463 			*oldlenp = ARG_MAX;	/* XXX XXX XXX */
2464 		error = 0;
2465 		mutex_exit(p->p_lock);
2466 		goto out_locked;
2467 	}
2468 
2469 	/*
2470 	 * Zombies don't have a stack, so we can't read their psstrings.
2471 	 * System processes also don't have a user stack.
2472 	 */
2473 	if (P_ZOMBIE(p) || (p->p_flag & PK_SYSTEM) != 0) {
2474 		error = EINVAL;
2475 		mutex_exit(p->p_lock);
2476 		goto out_locked;
2477 	}
2478 
2479 	/*
2480 	 * Lock the process down in memory.
2481 	 */
2482 	psstr_addr = (vaddr_t)p->p_psstr;
2483 	if (type == KERN_PROC_ARGV || type == KERN_PROC_NARGV) {
2484 		offsetn = p->p_psnargv;
2485 		offsetv = p->p_psargv;
2486 	} else {
2487 		offsetn = p->p_psnenv;
2488 		offsetv = p->p_psenv;
2489 	}
2490 	vmspace = p->p_vmspace;
2491 	uvmspace_addref(vmspace);
2492 	mutex_exit(p->p_lock);
2493 	mutex_exit(proc_lock);
2494 
2495 	/*
2496 	 * Allocate a temporary buffer to hold the arguments.
2497 	 */
2498 	arg = kmem_alloc(PAGE_SIZE, KM_SLEEP);
2499 
2500 	/*
2501 	 * Read in the ps_strings structure.
2502 	 */
2503 	aiov.iov_base = &pss;
2504 	aiov.iov_len = sizeof(pss);
2505 	auio.uio_iov = &aiov;
2506 	auio.uio_iovcnt = 1;
2507 	auio.uio_offset = psstr_addr;
2508 	auio.uio_resid = sizeof(pss);
2509 	auio.uio_rw = UIO_READ;
2510 	UIO_SETUP_SYSSPACE(&auio);
2511 	error = uvm_io(&vmspace->vm_map, &auio);
2512 	if (error)
2513 		goto done;
2514 
2515 	memcpy(&nargv, (char *)&pss + offsetn, sizeof(nargv));
2516 	if (type == KERN_PROC_NARGV || type == KERN_PROC_NENV) {
2517 		error = dcopyout(l, &nargv, oldp, sizeof(nargv));
2518 		*oldlenp = sizeof(nargv);
2519 		goto done;
2520 	}
2521 	/*
2522 	 * Now read the address of the argument vector.
2523 	 */
2524 	switch (type) {
2525 	case KERN_PROC_ARGV:
2526 		/* FALLTHROUGH */
2527 	case KERN_PROC_ENV:
2528 		memcpy(&tmp, (char *)&pss + offsetv, sizeof(tmp));
2529 		break;
2530 	default:
2531 		error = EINVAL;
2532 		goto done;
2533 	}
2534 
2535 #ifdef COMPAT_NETBSD32
2536 	if (p->p_flag & PK_32)
2537 		len = sizeof(netbsd32_charp) * nargv;
2538 	else
2539 #endif
2540 		len = sizeof(char *) * nargv;
2541 
2542 	if ((argvlen = len) != 0)
2543 		argv = kmem_alloc(len, KM_SLEEP);
2544 
2545 	aiov.iov_base = argv;
2546 	aiov.iov_len = len;
2547 	auio.uio_iov = &aiov;
2548 	auio.uio_iovcnt = 1;
2549 	auio.uio_offset = (off_t)(unsigned long)tmp;
2550 	auio.uio_resid = len;
2551 	auio.uio_rw = UIO_READ;
2552 	UIO_SETUP_SYSSPACE(&auio);
2553 	error = uvm_io(&vmspace->vm_map, &auio);
2554 	if (error)
2555 		goto done;
2556 
2557 	/*
2558 	 * Now copy each string.
2559 	 */
2560 	len = 0; /* bytes written to user buffer */
2561 	for (i = 0; i < nargv; i++) {
2562 		int finished = 0;
2563 		vaddr_t base;
2564 		size_t xlen;
2565 		int j;
2566 
2567 #ifdef COMPAT_NETBSD32
2568 		if (p->p_flag & PK_32) {
2569 			netbsd32_charp *argv32;
2570 
2571 			argv32 = (netbsd32_charp *)argv;
2572 			base = (vaddr_t)NETBSD32PTR64(argv32[i]);
2573 		} else
2574 #endif
2575 			base = (vaddr_t)argv[i];
2576 
2577 		/*
2578 		 * The program has messed around with its arguments,
2579 		 * possibly deleting some, and replacing them with
2580 		 * NULL's. Treat this as the last argument and not
2581 		 * a failure.
2582 		 */
2583 		if (base == 0)
2584 			break;
2585 
2586 		while (!finished) {
2587 			xlen = PAGE_SIZE - (base & PAGE_MASK);
2588 
2589 			aiov.iov_base = arg;
2590 			aiov.iov_len = PAGE_SIZE;
2591 			auio.uio_iov = &aiov;
2592 			auio.uio_iovcnt = 1;
2593 			auio.uio_offset = base;
2594 			auio.uio_resid = xlen;
2595 			auio.uio_rw = UIO_READ;
2596 			UIO_SETUP_SYSSPACE(&auio);
2597 			error = uvm_io(&vmspace->vm_map, &auio);
2598 			if (error)
2599 				goto done;
2600 
2601 			/* Look for the end of the string */
2602 			for (j = 0; j < xlen; j++) {
2603 				if (arg[j] == '\0') {
2604 					xlen = j + 1;
2605 					finished = 1;
2606 					break;
2607 				}
2608 			}
2609 
2610 			/* Check for user buffer overflow */
2611 			if (len + xlen > *oldlenp) {
2612 				finished = 1;
2613 				if (len > *oldlenp)
2614 					xlen = 0;
2615 				else
2616 					xlen = *oldlenp - len;
2617 			}
2618 
2619 			/* Copyout the page */
2620 			error = dcopyout(l, arg, (char *)oldp + len, xlen);
2621 			if (error)
2622 				goto done;
2623 
2624 			len += xlen;
2625 			base += xlen;
2626 		}
2627 	}
2628 	*oldlenp = len;
2629 
2630 done:
2631 	if (argvlen != 0)
2632 		kmem_free(argv, argvlen);
2633 	uvmspace_free(vmspace);
2634 	kmem_free(arg, PAGE_SIZE);
2635 	sysctl_relock();
2636 	return error;
2637 
2638 out_locked:
2639 	mutex_exit(proc_lock);
2640 	sysctl_relock();
2641 	return error;
2642 }
2643 
2644 static int
2645 sysctl_security_setidcore(SYSCTLFN_ARGS)
2646 {
2647 	int newsize, error;
2648 	struct sysctlnode node;
2649 
2650 	node = *rnode;
2651 	node.sysctl_data = &newsize;
2652 	newsize = *(int *)rnode->sysctl_data;
2653 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
2654 	if (error || newp == NULL)
2655 		return error;
2656 
2657 	if (kauth_authorize_system(l->l_cred, KAUTH_SYSTEM_SETIDCORE,
2658 	    0, NULL, NULL, NULL))
2659 		return (EPERM);
2660 
2661 	*(int *)rnode->sysctl_data = newsize;
2662 
2663 	return 0;
2664 }
2665 
2666 static int
2667 sysctl_security_setidcorename(SYSCTLFN_ARGS)
2668 {
2669 	int error;
2670 	char *newsetidcorename;
2671 	struct sysctlnode node;
2672 
2673 	newsetidcorename = PNBUF_GET();
2674 	node = *rnode;
2675 	node.sysctl_data = newsetidcorename;
2676 	memcpy(node.sysctl_data, rnode->sysctl_data, MAXPATHLEN);
2677 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
2678 	if (error || newp == NULL) {
2679 		goto out;
2680 	}
2681 	if (kauth_authorize_system(l->l_cred, KAUTH_SYSTEM_SETIDCORE,
2682 	    0, NULL, NULL, NULL)) {
2683 		error = EPERM;
2684 		goto out;
2685 	}
2686 	if (strlen(newsetidcorename) == 0) {
2687 		error = EINVAL;
2688 		goto out;
2689 	}
2690 	memcpy(rnode->sysctl_data, node.sysctl_data, MAXPATHLEN);
2691 out:
2692 	PNBUF_PUT(newsetidcorename);
2693 	return error;
2694 }
2695 
2696 /*
2697  * sysctl helper routine for kern.cp_id node. Maps cpus to their
2698  * cpuids.
2699  */
2700 static int
2701 sysctl_kern_cpid(SYSCTLFN_ARGS)
2702 {
2703 	struct sysctlnode node = *rnode;
2704 	uint64_t *cp_id = NULL;
2705 	int error, n = ncpu;
2706 	struct cpu_info *ci;
2707 	CPU_INFO_ITERATOR cii;
2708 
2709 	/*
2710 	 * Here you may either retrieve a single cpu id or the whole
2711 	 * set. The size you get back when probing depends on what
2712 	 * you ask for.
2713 	 */
2714 	switch (namelen) {
2715 	case 0:
2716 		node.sysctl_size = n * sizeof(uint64_t);
2717 		n = -2; /* ALL */
2718 		break;
2719 	case 1:
2720 		if (name[0] < 0 || name[0] >= n)
2721 			return (ENOENT); /* ENOSUCHPROCESSOR */
2722 		node.sysctl_size = sizeof(uint64_t);
2723 		n = name[0];
2724 		/*
2725 		 * adjust these so that sysctl_lookup() will be happy
2726 		 */
2727 		name++;
2728 		namelen--;
2729 		break;
2730 	default:
2731 		return (EINVAL);
2732 	}
2733 
2734 	cp_id = kmem_alloc(node.sysctl_size, KM_SLEEP);
2735 	if (cp_id == NULL)
2736 		return (ENOMEM);
2737 	node.sysctl_data = cp_id;
2738 	memset(cp_id, 0, node.sysctl_size);
2739 
2740 	for (CPU_INFO_FOREACH(cii, ci)) {
2741 		if (n <= 0)
2742 			cp_id[0] = cpu_index(ci);
2743 		/*
2744 		 * if a specific processor was requested and we just
2745 		 * did it, we're done here
2746 		 */
2747 		if (n == 0)
2748 			break;
2749 		/*
2750 		 * if doing "all", skip to next cp_id slot for next processor
2751 		 */
2752 		if (n == -2)
2753 			cp_id++;
2754 		/*
2755 		 * if we're doing a specific processor, we're one
2756 		 * processor closer
2757 		 */
2758 		if (n > 0)
2759 			n--;
2760 	}
2761 
2762 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
2763 	kmem_free(node.sysctl_data, node.sysctl_size);
2764 	return (error);
2765 }
2766 
2767 /*
2768  * sysctl helper routine for hw.usermem and hw.usermem64. Values are
2769  * calculate on the fly taking into account integer overflow and the
2770  * current wired count.
2771  */
2772 static int
2773 sysctl_hw_usermem(SYSCTLFN_ARGS)
2774 {
2775 	u_int ui;
2776 	u_quad_t uq;
2777 	struct sysctlnode node;
2778 
2779 	node = *rnode;
2780 	switch (rnode->sysctl_num) {
2781 	    case HW_USERMEM:
2782 		if ((ui = physmem - uvmexp.wired) > (UINT_MAX / PAGE_SIZE))
2783 			ui = UINT_MAX;
2784 		else
2785 			ui *= PAGE_SIZE;
2786 		node.sysctl_data = &ui;
2787 		break;
2788 	case HW_USERMEM64:
2789 		uq = (u_quad_t)(physmem - uvmexp.wired) * PAGE_SIZE;
2790 		node.sysctl_data = &uq;
2791 		break;
2792 	default:
2793 		return (EINVAL);
2794 	}
2795 
2796 	return (sysctl_lookup(SYSCTLFN_CALL(&node)));
2797 }
2798 
2799 /*
2800  * sysctl helper routine for kern.cnmagic node. Pulls the old value
2801  * out, encoded, and stuffs the new value in for decoding.
2802  */
2803 static int
2804 sysctl_hw_cnmagic(SYSCTLFN_ARGS)
2805 {
2806 	char magic[CNS_LEN];
2807 	int error;
2808 	struct sysctlnode node;
2809 
2810 	if (oldp)
2811 		cn_get_magic(magic, CNS_LEN);
2812 	node = *rnode;
2813 	node.sysctl_data = &magic[0];
2814 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
2815 	if (error || newp == NULL)
2816 		return (error);
2817 
2818 	return (cn_set_magic(magic));
2819 }
2820 
2821 /*
2822  * ********************************************************************
2823  * section 3: public helper routines that are used for more than one
2824  * node
2825  * ********************************************************************
2826  */
2827 
2828 /*
2829  * sysctl helper routine for the kern.root_device node and some ports'
2830  * machdep.root_device nodes.
2831  */
2832 int
2833 sysctl_root_device(SYSCTLFN_ARGS)
2834 {
2835 	struct sysctlnode node;
2836 
2837 	node = *rnode;
2838 	node.sysctl_data = root_device->dv_xname;
2839 	node.sysctl_size = strlen(device_xname(root_device)) + 1;
2840 	return (sysctl_lookup(SYSCTLFN_CALL(&node)));
2841 }
2842 
2843 /*
2844  * sysctl helper routine for kern.consdev, dependent on the current
2845  * state of the console. Also used for machdep.console_device on some
2846  * ports.
2847  */
2848 int
2849 sysctl_consdev(SYSCTLFN_ARGS)
2850 {
2851 	dev_t consdev;
2852 	struct sysctlnode node;
2853 
2854 	if (cn_tab != NULL)
2855 		consdev = cn_tab->cn_dev;
2856 	else
2857 		consdev = NODEV;
2858 	node = *rnode;
2859 	node.sysctl_data = &consdev;
2860 	node.sysctl_size = sizeof(consdev);
2861 	return (sysctl_lookup(SYSCTLFN_CALL(&node)));
2862 }
2863 
2864 /*
2865  * ********************************************************************
2866  * section 4: support for some helpers
2867  * ********************************************************************
2868  */
2869 
2870 /*
2871  * Fill in a kinfo_proc2 structure for the specified process.
2872  */
2873 static void
2874 fill_kproc2(struct proc *p, struct kinfo_proc2 *ki, bool zombie)
2875 {
2876 	struct tty *tp;
2877 	struct lwp *l, *l2;
2878 	struct timeval ut, st, rt;
2879 	sigset_t ss1, ss2;
2880 	struct rusage ru;
2881 	struct vmspace *vm;
2882 
2883 	KASSERT(mutex_owned(proc_lock));
2884 	KASSERT(mutex_owned(p->p_lock));
2885 
2886 	sigemptyset(&ss1);
2887 	sigemptyset(&ss2);
2888 	memset(ki, 0, sizeof(*ki));
2889 
2890 	ki->p_paddr = PTRTOUINT64(p);
2891 	ki->p_fd = PTRTOUINT64(p->p_fd);
2892 	ki->p_cwdi = PTRTOUINT64(p->p_cwdi);
2893 	ki->p_stats = PTRTOUINT64(p->p_stats);
2894 	ki->p_limit = PTRTOUINT64(p->p_limit);
2895 	ki->p_vmspace = PTRTOUINT64(p->p_vmspace);
2896 	ki->p_sigacts = PTRTOUINT64(p->p_sigacts);
2897 	ki->p_sess = PTRTOUINT64(p->p_session);
2898 	ki->p_tsess = 0;	/* may be changed if controlling tty below */
2899 	ki->p_ru = PTRTOUINT64(&p->p_stats->p_ru);
2900 	ki->p_eflag = 0;
2901 	ki->p_exitsig = p->p_exitsig;
2902 	ki->p_flag = sysctl_map_flags(sysctl_flagmap, p->p_flag);
2903 	ki->p_flag |= sysctl_map_flags(sysctl_sflagmap, p->p_sflag);
2904 	ki->p_flag |= sysctl_map_flags(sysctl_slflagmap, p->p_slflag);
2905 	ki->p_flag |= sysctl_map_flags(sysctl_lflagmap, p->p_lflag);
2906 	ki->p_flag |= sysctl_map_flags(sysctl_stflagmap, p->p_stflag);
2907 	ki->p_pid = p->p_pid;
2908 	if (p->p_pptr)
2909 		ki->p_ppid = p->p_pptr->p_pid;
2910 	else
2911 		ki->p_ppid = 0;
2912 	ki->p_uid = kauth_cred_geteuid(p->p_cred);
2913 	ki->p_ruid = kauth_cred_getuid(p->p_cred);
2914 	ki->p_gid = kauth_cred_getegid(p->p_cred);
2915 	ki->p_rgid = kauth_cred_getgid(p->p_cred);
2916 	ki->p_svuid = kauth_cred_getsvuid(p->p_cred);
2917 	ki->p_svgid = kauth_cred_getsvgid(p->p_cred);
2918 	ki->p_ngroups = kauth_cred_ngroups(p->p_cred);
2919 	kauth_cred_getgroups(p->p_cred, ki->p_groups,
2920 	    min(ki->p_ngroups, sizeof(ki->p_groups) / sizeof(ki->p_groups[0])),
2921 	    UIO_SYSSPACE);
2922 
2923 	ki->p_uticks = p->p_uticks;
2924 	ki->p_sticks = p->p_sticks;
2925 	ki->p_iticks = p->p_iticks;
2926 	ki->p_tpgid = NO_PGID;	/* may be changed if controlling tty below */
2927 	ki->p_tracep = PTRTOUINT64(p->p_tracep);
2928 	ki->p_traceflag = p->p_traceflag;
2929 
2930 	memcpy(&ki->p_sigignore, &p->p_sigctx.ps_sigignore,sizeof(ki_sigset_t));
2931 	memcpy(&ki->p_sigcatch, &p->p_sigctx.ps_sigcatch, sizeof(ki_sigset_t));
2932 
2933 	ki->p_cpticks = 0;
2934 	ki->p_pctcpu = p->p_pctcpu;
2935 	ki->p_estcpu = 0;
2936 	ki->p_stat = p->p_stat; /* Will likely be overridden by LWP status */
2937 	ki->p_realstat = p->p_stat;
2938 	ki->p_nice = p->p_nice;
2939 	ki->p_xstat = p->p_xstat;
2940 	ki->p_acflag = p->p_acflag;
2941 
2942 	strncpy(ki->p_comm, p->p_comm,
2943 	    min(sizeof(ki->p_comm), sizeof(p->p_comm)));
2944 	strncpy(ki->p_ename, p->p_emul->e_name, sizeof(ki->p_ename));
2945 
2946 	ki->p_nlwps = p->p_nlwps;
2947 	ki->p_realflag = ki->p_flag;
2948 
2949 	if (p->p_stat != SIDL && !P_ZOMBIE(p) && !zombie) {
2950 		vm = p->p_vmspace;
2951 		ki->p_vm_rssize = vm_resident_count(vm);
2952 		ki->p_vm_tsize = vm->vm_tsize;
2953 		ki->p_vm_dsize = vm->vm_dsize;
2954 		ki->p_vm_ssize = vm->vm_ssize;
2955 
2956 		/* Pick the primary (first) LWP */
2957 		l = LIST_FIRST(&p->p_lwps);
2958 		KASSERT(l != NULL);
2959 		lwp_lock(l);
2960 		ki->p_nrlwps = p->p_nrlwps;
2961 		ki->p_forw = 0;
2962 		ki->p_back = 0;
2963 		ki->p_addr = PTRTOUINT64(l->l_addr);
2964 		ki->p_stat = l->l_stat;
2965 		ki->p_flag |= sysctl_map_flags(sysctl_lwpflagmap, l->l_flag);
2966 		ki->p_swtime = l->l_swtime;
2967 		ki->p_slptime = l->l_slptime;
2968 		if (l->l_stat == LSONPROC)
2969 			ki->p_schedflags = l->l_cpu->ci_schedstate.spc_flags;
2970 		else
2971 			ki->p_schedflags = 0;
2972 		ki->p_holdcnt = l->l_holdcnt;
2973 		ki->p_priority = lwp_eprio(l);
2974 		ki->p_usrpri = l->l_priority;
2975 		if (l->l_wchan)
2976 			strncpy(ki->p_wmesg, l->l_wmesg, sizeof(ki->p_wmesg));
2977 		ki->p_wchan = PTRTOUINT64(l->l_wchan);
2978 		ki->p_cpuid = cpu_index(l->l_cpu);
2979 		lwp_unlock(l);
2980 		LIST_FOREACH(l, &p->p_lwps, l_sibling) {
2981 			/* This is hardly correct, but... */
2982 			sigplusset(&l->l_sigpend.sp_set, &ss1);
2983 			sigplusset(&l->l_sigmask, &ss2);
2984 			ki->p_cpticks += l->l_cpticks;
2985 			ki->p_pctcpu += l->l_pctcpu;
2986 			ki->p_estcpu += l->l_estcpu;
2987 		}
2988 	}
2989 	sigplusset(&p->p_sigpend.sp_set, &ss2);
2990 	memcpy(&ki->p_siglist, &ss1, sizeof(ki_sigset_t));
2991 	memcpy(&ki->p_sigmask, &ss2, sizeof(ki_sigset_t));
2992 
2993 	if (p->p_session != NULL) {
2994 		ki->p_sid = p->p_session->s_sid;
2995 		ki->p__pgid = p->p_pgrp->pg_id;
2996 		if (p->p_session->s_ttyvp)
2997 			ki->p_eflag |= EPROC_CTTY;
2998 		if (SESS_LEADER(p))
2999 			ki->p_eflag |= EPROC_SLEADER;
3000 		strncpy(ki->p_login, p->p_session->s_login,
3001 		    min(sizeof ki->p_login - 1, sizeof p->p_session->s_login));
3002 		ki->p_jobc = p->p_pgrp->pg_jobc;
3003 		if ((p->p_lflag & PL_CONTROLT) && (tp = p->p_session->s_ttyp)) {
3004 			ki->p_tdev = tp->t_dev;
3005 			ki->p_tpgid = tp->t_pgrp ? tp->t_pgrp->pg_id : NO_PGID;
3006 			ki->p_tsess = PTRTOUINT64(tp->t_session);
3007 		} else {
3008 			ki->p_tdev = NODEV;
3009 		}
3010 	}
3011 
3012 	if (!P_ZOMBIE(p) && !zombie) {
3013 		ki->p_uvalid = 1;
3014 		ki->p_ustart_sec = p->p_stats->p_start.tv_sec;
3015 		ki->p_ustart_usec = p->p_stats->p_start.tv_usec;
3016 
3017 		calcru(p, &ut, &st, NULL, &rt);
3018 		ki->p_rtime_sec = rt.tv_sec;
3019 		ki->p_rtime_usec = rt.tv_usec;
3020 		ki->p_uutime_sec = ut.tv_sec;
3021 		ki->p_uutime_usec = ut.tv_usec;
3022 		ki->p_ustime_sec = st.tv_sec;
3023 		ki->p_ustime_usec = st.tv_usec;
3024 
3025 		memcpy(&ru, &p->p_stats->p_ru, sizeof(ru));
3026 		ki->p_uru_nvcsw = 0;
3027 		ki->p_uru_nivcsw = 0;
3028 		LIST_FOREACH(l2, &p->p_lwps, l_sibling) {
3029 			ki->p_uru_nvcsw += (l2->l_ncsw - l2->l_nivcsw);
3030 			ki->p_uru_nivcsw += l2->l_nivcsw;
3031 			ruadd(&ru, &l2->l_ru);
3032 		}
3033 		ki->p_uru_maxrss = ru.ru_maxrss;
3034 		ki->p_uru_ixrss = ru.ru_ixrss;
3035 		ki->p_uru_idrss = ru.ru_idrss;
3036 		ki->p_uru_isrss = ru.ru_isrss;
3037 		ki->p_uru_minflt = ru.ru_minflt;
3038 		ki->p_uru_majflt = ru.ru_majflt;
3039 		ki->p_uru_nswap = ru.ru_nswap;
3040 		ki->p_uru_inblock = ru.ru_inblock;
3041 		ki->p_uru_oublock = ru.ru_oublock;
3042 		ki->p_uru_msgsnd = ru.ru_msgsnd;
3043 		ki->p_uru_msgrcv = ru.ru_msgrcv;
3044 		ki->p_uru_nsignals = ru.ru_nsignals;
3045 
3046 		timeradd(&p->p_stats->p_cru.ru_utime,
3047 			 &p->p_stats->p_cru.ru_stime, &ut);
3048 		ki->p_uctime_sec = ut.tv_sec;
3049 		ki->p_uctime_usec = ut.tv_usec;
3050 	}
3051 }
3052 
3053 /*
3054  * Fill in a kinfo_lwp structure for the specified lwp.
3055  */
3056 static void
3057 fill_lwp(struct lwp *l, struct kinfo_lwp *kl)
3058 {
3059 	struct proc *p = l->l_proc;
3060 	struct timeval tv;
3061 
3062 	KASSERT(lwp_locked(l, NULL));
3063 
3064 	kl->l_forw = 0;
3065 	kl->l_back = 0;
3066 	kl->l_laddr = PTRTOUINT64(l);
3067 	kl->l_addr = PTRTOUINT64(l->l_addr);
3068 	kl->l_stat = l->l_stat;
3069 	kl->l_lid = l->l_lid;
3070 	kl->l_flag = sysctl_map_flags(sysctl_lwpprflagmap, l->l_prflag);
3071 	kl->l_flag |= sysctl_map_flags(sysctl_lwpflagmap, l->l_flag);
3072 
3073 	kl->l_swtime = l->l_swtime;
3074 	kl->l_slptime = l->l_slptime;
3075 	if (l->l_stat == LSONPROC)
3076 		kl->l_schedflags = l->l_cpu->ci_schedstate.spc_flags;
3077 	else
3078 		kl->l_schedflags = 0;
3079 	kl->l_holdcnt = l->l_holdcnt;
3080 	kl->l_priority = lwp_eprio(l);
3081 	kl->l_usrpri = l->l_priority;
3082 	if (l->l_wchan)
3083 		strncpy(kl->l_wmesg, l->l_wmesg, sizeof(kl->l_wmesg));
3084 	kl->l_wchan = PTRTOUINT64(l->l_wchan);
3085 	kl->l_cpuid = cpu_index(l->l_cpu);
3086 	bintime2timeval(&l->l_rtime, &tv);
3087 	kl->l_rtime_sec = tv.tv_sec;
3088 	kl->l_rtime_usec = tv.tv_usec;
3089 	kl->l_cpticks = l->l_cpticks;
3090 	kl->l_pctcpu = l->l_pctcpu;
3091 	kl->l_pid = p->p_pid;
3092 	if (l->l_name == NULL)
3093 		kl->l_name[0] = '\0';
3094 	else
3095 		strlcpy(kl->l_name, l->l_name, sizeof(kl->l_name));
3096 }
3097 
3098 /*
3099  * Fill in an eproc structure for the specified process.
3100  */
3101 void
3102 fill_eproc(struct proc *p, struct eproc *ep, bool zombie)
3103 {
3104 	struct tty *tp;
3105 	struct lwp *l;
3106 
3107 	KASSERT(mutex_owned(proc_lock));
3108 	KASSERT(mutex_owned(p->p_lock));
3109 
3110 	memset(ep, 0, sizeof(*ep));
3111 
3112 	ep->e_paddr = p;
3113 	ep->e_sess = p->p_session;
3114 	if (p->p_cred) {
3115 		kauth_cred_topcred(p->p_cred, &ep->e_pcred);
3116 		kauth_cred_toucred(p->p_cred, &ep->e_ucred);
3117 	}
3118 	if (p->p_stat != SIDL && !P_ZOMBIE(p) && !zombie) {
3119 		struct vmspace *vm = p->p_vmspace;
3120 
3121 		ep->e_vm.vm_rssize = vm_resident_count(vm);
3122 		ep->e_vm.vm_tsize = vm->vm_tsize;
3123 		ep->e_vm.vm_dsize = vm->vm_dsize;
3124 		ep->e_vm.vm_ssize = vm->vm_ssize;
3125 
3126 		/* Pick the primary (first) LWP */
3127 		l = LIST_FIRST(&p->p_lwps);
3128 		KASSERT(l != NULL);
3129 		lwp_lock(l);
3130 		if (l->l_wchan)
3131 			strncpy(ep->e_wmesg, l->l_wmesg, WMESGLEN);
3132 		lwp_unlock(l);
3133 	}
3134 	if (p->p_pptr)
3135 		ep->e_ppid = p->p_pptr->p_pid;
3136 	if (p->p_pgrp && p->p_session) {
3137 		ep->e_pgid = p->p_pgrp->pg_id;
3138 		ep->e_jobc = p->p_pgrp->pg_jobc;
3139 		ep->e_sid = p->p_session->s_sid;
3140 		if ((p->p_lflag & PL_CONTROLT) &&
3141 		    (tp = ep->e_sess->s_ttyp)) {
3142 			ep->e_tdev = tp->t_dev;
3143 			ep->e_tpgid = tp->t_pgrp ? tp->t_pgrp->pg_id : NO_PGID;
3144 			ep->e_tsess = tp->t_session;
3145 		} else
3146 			ep->e_tdev = NODEV;
3147 		ep->e_flag = ep->e_sess->s_ttyvp ? EPROC_CTTY : 0;
3148 		if (SESS_LEADER(p))
3149 			ep->e_flag |= EPROC_SLEADER;
3150 		strncpy(ep->e_login, ep->e_sess->s_login, MAXLOGNAME);
3151 	}
3152 	ep->e_xsize = ep->e_xrssize = 0;
3153 	ep->e_xccount = ep->e_xswrss = 0;
3154 }
3155 
3156 u_int
3157 sysctl_map_flags(const u_int *map, u_int word)
3158 {
3159 	u_int rv;
3160 
3161 	for (rv = 0; *map != 0; map += 2)
3162 		if ((word & map[0]) != 0)
3163 			rv |= map[1];
3164 
3165 	return rv;
3166 }
3167