init_sysctl.c revision 1.224
1/*	$NetBSD: init_sysctl.c,v 1.224 2020/01/18 14:40:03 skrll Exp $ */
2
3/*-
4 * Copyright (c) 2003, 2007, 2008, 2009 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 1.224 2020/01/18 14:40:03 skrll Exp $");
34
35#include "opt_sysv.h"
36#include "opt_compat_netbsd.h"
37#include "opt_modular.h"
38#include "opt_gprof.h"
39#include "pty.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/cprng.h>
51#include <sys/vnode_impl.h>	/* For vfs_drainvnodes(). */
52#include <sys/mount.h>
53#include <sys/namei.h>
54#include <dev/cons.h>
55#include <sys/socketvar.h>
56#include <sys/file.h>
57#include <sys/filedesc.h>
58#include <sys/tty.h>
59#include <sys/kmem.h>
60#include <sys/reboot.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 <sys/cpu.h>
71
72int security_setidcore_dump;
73char security_setidcore_path[MAXPATHLEN] = "/var/crash/%n.core";
74uid_t security_setidcore_owner = 0;
75gid_t security_setidcore_group = 0;
76mode_t security_setidcore_mode = (S_IRUSR|S_IWUSR);
77
78/*
79 * Current status of SysV IPC capability.  Initially, these are
80 * 0 if the capability is not built-in to the kernel, but can
81 * be updated if the appropriate kernel module is (auto)loaded.
82 */
83
84int kern_has_sysvmsg = 0;
85int kern_has_sysvshm = 0;
86int kern_has_sysvsem = 0;
87
88static const u_int sysctl_lwpprflagmap[] = {
89	LPR_DETACHED, L_DETACHED,
90	0
91};
92
93/*
94 * try over estimating by 5 procs/lwps
95 */
96#define KERN_LWPSLOP	(5 * sizeof(struct kinfo_lwp))
97
98static int dcopyout(struct lwp *, const void *, void *, size_t);
99
100static int
101dcopyout(struct lwp *l, const void *kaddr, void *uaddr, size_t len)
102{
103	int error;
104
105	error = copyout(kaddr, uaddr, len);
106	ktrmibio(-1, UIO_READ, uaddr, len, error);
107
108	return error;
109}
110
111static int sysctl_kern_maxvnodes(SYSCTLFN_PROTO);
112static int sysctl_kern_messages(SYSCTLFN_PROTO);
113static int sysctl_kern_boottime(SYSCTLFN_PROTO);
114static int sysctl_kern_rtc_offset(SYSCTLFN_PROTO);
115static int sysctl_kern_maxproc(SYSCTLFN_PROTO);
116static int sysctl_kern_hostid(SYSCTLFN_PROTO);
117static int sysctl_kern_defcorename(SYSCTLFN_PROTO);
118static int sysctl_kern_cptime(SYSCTLFN_PROTO);
119#if NPTY > 0
120static int sysctl_kern_maxptys(SYSCTLFN_PROTO);
121#endif /* NPTY > 0 */
122static int sysctl_kern_lwp(SYSCTLFN_PROTO);
123static int sysctl_kern_forkfsleep(SYSCTLFN_PROTO);
124static int sysctl_kern_root_partition(SYSCTLFN_PROTO);
125static int sysctl_kern_drivers(SYSCTLFN_PROTO);
126static int sysctl_security_setidcore(SYSCTLFN_PROTO);
127static int sysctl_security_setidcorename(SYSCTLFN_PROTO);
128static int sysctl_kern_cpid(SYSCTLFN_PROTO);
129static int sysctl_hw_usermem(SYSCTLFN_PROTO);
130static int sysctl_hw_cnmagic(SYSCTLFN_PROTO);
131
132static void fill_lwp(struct lwp *l, struct kinfo_lwp *kl);
133
134/*
135 * ********************************************************************
136 * section 1: setup routines
137 * ********************************************************************
138 * These functions are stuffed into a link set for sysctl setup
139 * functions. They're never called or referenced from anywhere else.
140 * ********************************************************************
141 */
142
143/*
144 * this setup routine is a replacement for kern_sysctl()
145 */
146SYSCTL_SETUP(sysctl_kern_setup, "sysctl kern subtree setup")
147{
148	extern int kern_logsigexit;	/* defined in kern/kern_sig.c */
149	extern fixpt_t ccpu;		/* defined in kern/kern_synch.c */
150	extern int dumponpanic;		/* defined in kern/subr_prf.c */
151	const struct sysctlnode *rnode;
152
153	sysctl_createv(clog, 0, NULL, NULL,
154		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
155		       CTLTYPE_INT, "maxvnodes",
156		       SYSCTL_DESCR("Maximum number of vnodes"),
157		       sysctl_kern_maxvnodes, 0, NULL, 0,
158		       CTL_KERN, KERN_MAXVNODES, CTL_EOL);
159	sysctl_createv(clog, 0, NULL, NULL,
160		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
161		       CTLTYPE_INT, "maxproc",
162		       SYSCTL_DESCR("Maximum number of simultaneous processes"),
163		       sysctl_kern_maxproc, 0, NULL, 0,
164		       CTL_KERN, KERN_MAXPROC, CTL_EOL);
165	sysctl_createv(clog, 0, NULL, NULL,
166		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
167		       CTLTYPE_INT, "maxfiles",
168		       SYSCTL_DESCR("Maximum number of open files"),
169		       NULL, 0, &maxfiles, 0,
170		       CTL_KERN, KERN_MAXFILES, CTL_EOL);
171	sysctl_createv(clog, 0, NULL, NULL,
172		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
173		       CTLTYPE_INT, "argmax",
174		       SYSCTL_DESCR("Maximum number of bytes of arguments to "
175				    "execve(2)"),
176		       NULL, ARG_MAX, NULL, 0,
177		       CTL_KERN, KERN_ARGMAX, CTL_EOL);
178	sysctl_createv(clog, 0, NULL, NULL,
179		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE|CTLFLAG_HEX,
180		       CTLTYPE_INT, "hostid",
181		       SYSCTL_DESCR("System host ID number"),
182		       sysctl_kern_hostid, 0, NULL, 0,
183		       CTL_KERN, KERN_HOSTID, CTL_EOL);
184	sysctl_createv(clog, 0, NULL, NULL,
185		       CTLFLAG_PERMANENT,
186		       CTLTYPE_STRUCT, "vnode",
187		       SYSCTL_DESCR("System vnode table"),
188		       sysctl_kern_vnode, 0, NULL, 0,
189		       CTL_KERN, KERN_VNODE, CTL_EOL);
190#ifndef GPROF
191	sysctl_createv(clog, 0, NULL, NULL,
192		       CTLFLAG_PERMANENT,
193		       CTLTYPE_NODE, "profiling",
194		       SYSCTL_DESCR("Profiling information (not available)"),
195		       sysctl_notavail, 0, NULL, 0,
196		       CTL_KERN, KERN_PROF, CTL_EOL);
197#endif
198	sysctl_createv(clog, 0, NULL, NULL,
199		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
200		       CTLTYPE_INT, "posix1version",
201		       SYSCTL_DESCR("Version of ISO/IEC 9945 (POSIX 1003.1) "
202				    "with which the operating system attempts "
203				    "to comply"),
204		       NULL, _POSIX_VERSION, NULL, 0,
205		       CTL_KERN, KERN_POSIX1, CTL_EOL);
206	sysctl_createv(clog, 0, NULL, NULL,
207		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
208		       CTLTYPE_INT, "ngroups",
209		       SYSCTL_DESCR("Maximum number of supplemental groups"),
210		       NULL, NGROUPS_MAX, NULL, 0,
211		       CTL_KERN, KERN_NGROUPS, CTL_EOL);
212	sysctl_createv(clog, 0, NULL, NULL,
213		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
214		       CTLTYPE_INT, "job_control",
215		       SYSCTL_DESCR("Whether job control is available"),
216		       NULL, 1, NULL, 0,
217		       CTL_KERN, KERN_JOB_CONTROL, CTL_EOL);
218	sysctl_createv(clog, 0, NULL, NULL,
219		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
220		       CTLTYPE_INT, "saved_ids",
221		       SYSCTL_DESCR("Whether POSIX saved set-group/user ID is "
222				    "available"), NULL,
223#ifdef _POSIX_SAVED_IDS
224		       1,
225#else /* _POSIX_SAVED_IDS */
226		       0,
227#endif /* _POSIX_SAVED_IDS */
228		       NULL, 0, CTL_KERN, KERN_SAVED_IDS, CTL_EOL);
229	sysctl_createv(clog, 0, NULL, NULL,
230		       CTLFLAG_PERMANENT|CTLFLAG_HEX,
231		       CTLTYPE_INT, "boothowto",
232		       SYSCTL_DESCR("Flags from boot loader"),
233		       NULL, 0, &boothowto, sizeof(boothowto),
234		       CTL_KERN, CTL_CREATE, CTL_EOL);
235	sysctl_createv(clog, 0, NULL, NULL,
236		       CTLFLAG_PERMANENT,
237		       CTLTYPE_STRUCT, "boottime",
238		       SYSCTL_DESCR("System boot time"),
239		       sysctl_kern_boottime, 0, NULL, sizeof(struct timespec),
240		       CTL_KERN, KERN_BOOTTIME, CTL_EOL);
241	sysctl_createv(clog, 0, NULL, NULL,
242		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
243		       CTLTYPE_INT, "maxpartitions",
244		       SYSCTL_DESCR("Maximum number of partitions allowed per "
245				    "disk"),
246		       NULL, MAXPARTITIONS, NULL, 0,
247		       CTL_KERN, KERN_MAXPARTITIONS, CTL_EOL);
248	sysctl_createv(clog, 0, NULL, NULL,
249		       CTLFLAG_PERMANENT,
250		       CTLTYPE_STRUCT, "timex", NULL,
251		       sysctl_notavail, 0, NULL, 0,
252		       CTL_KERN, KERN_TIMEX, CTL_EOL);
253	sysctl_createv(clog, 0, NULL, NULL,
254		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
255		       CTLTYPE_INT, "rtc_offset",
256		       SYSCTL_DESCR("Offset of real time clock from UTC in "
257				    "minutes"),
258		       sysctl_kern_rtc_offset, 0, &rtc_offset, 0,
259		       CTL_KERN, KERN_RTC_OFFSET, CTL_EOL);
260	sysctl_createv(clog, 0, NULL, NULL,
261		       CTLFLAG_PERMANENT,
262		       CTLTYPE_STRING, "root_device",
263		       SYSCTL_DESCR("Name of the root device"),
264		       sysctl_root_device, 0, NULL, 0,
265		       CTL_KERN, KERN_ROOT_DEVICE, CTL_EOL);
266	sysctl_createv(clog, 0, NULL, NULL,
267		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
268		       CTLTYPE_INT, "fsync",
269		       SYSCTL_DESCR("Whether the POSIX 1003.1b File "
270				    "Synchronization Option is available on "
271				    "this system"),
272		       NULL, 1, NULL, 0,
273		       CTL_KERN, KERN_FSYNC, CTL_EOL);
274	sysctl_createv(clog, 0, NULL, NULL,
275		       CTLFLAG_PERMANENT,
276		       CTLTYPE_NODE, "ipc",
277		       SYSCTL_DESCR("SysV IPC options"),
278		       NULL, 0, NULL, 0,
279		       CTL_KERN, KERN_SYSVIPC, CTL_EOL);
280	sysctl_createv(clog, 0, NULL, NULL,
281		       CTLFLAG_PERMANENT|CTLFLAG_READONLY,
282		       CTLTYPE_INT, "sysvmsg",
283		       SYSCTL_DESCR("System V style message support available"),
284		       NULL, 0, &kern_has_sysvmsg, sizeof(int),
285		       CTL_KERN, KERN_SYSVIPC, KERN_SYSVIPC_MSG, CTL_EOL);
286	sysctl_createv(clog, 0, NULL, NULL,
287		       CTLFLAG_PERMANENT|CTLFLAG_READONLY,
288		       CTLTYPE_INT, "sysvsem",
289		       SYSCTL_DESCR("System V style semaphore support "
290				    "available"),
291		       NULL, 0, &kern_has_sysvsem, sizeof(int),
292		       CTL_KERN, KERN_SYSVIPC, KERN_SYSVIPC_SEM, CTL_EOL);
293	sysctl_createv(clog, 0, NULL, NULL,
294		       CTLFLAG_PERMANENT|CTLFLAG_READONLY,
295		       CTLTYPE_INT, "sysvshm",
296		       SYSCTL_DESCR("System V style shared memory support "
297				    "available"),
298		       NULL, 0, &kern_has_sysvshm, sizeof(int),
299		       CTL_KERN, KERN_SYSVIPC, KERN_SYSVIPC_SHM, CTL_EOL);
300	sysctl_createv(clog, 0, NULL, NULL,
301		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
302		       CTLTYPE_INT, "synchronized_io",
303		       SYSCTL_DESCR("Whether the POSIX 1003.1b Synchronized "
304				    "I/O Option is available on this system"),
305		       NULL, 1, NULL, 0,
306		       CTL_KERN, KERN_SYNCHRONIZED_IO, CTL_EOL);
307	sysctl_createv(clog, 0, NULL, NULL,
308		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
309		       CTLTYPE_INT, "iov_max",
310		       SYSCTL_DESCR("Maximum number of iovec structures per "
311				    "process"),
312		       NULL, IOV_MAX, NULL, 0,
313		       CTL_KERN, KERN_IOV_MAX, CTL_EOL);
314	sysctl_createv(clog, 0, NULL, NULL,
315		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
316		       CTLTYPE_INT, "mapped_files",
317		       SYSCTL_DESCR("Whether the POSIX 1003.1b Memory Mapped "
318				    "Files Option is available on this system"),
319		       NULL, 1, NULL, 0,
320		       CTL_KERN, KERN_MAPPED_FILES, CTL_EOL);
321	sysctl_createv(clog, 0, NULL, NULL,
322		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
323		       CTLTYPE_INT, "memlock",
324		       SYSCTL_DESCR("Whether the POSIX 1003.1b Process Memory "
325				    "Locking Option is available on this "
326				    "system"),
327		       NULL, 1, NULL, 0,
328		       CTL_KERN, KERN_MEMLOCK, CTL_EOL);
329	sysctl_createv(clog, 0, NULL, NULL,
330		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
331		       CTLTYPE_INT, "memlock_range",
332		       SYSCTL_DESCR("Whether the POSIX 1003.1b Range Memory "
333				    "Locking Option is available on this "
334				    "system"),
335		       NULL, 1, NULL, 0,
336		       CTL_KERN, KERN_MEMLOCK_RANGE, CTL_EOL);
337	sysctl_createv(clog, 0, NULL, NULL,
338		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
339		       CTLTYPE_INT, "memory_protection",
340		       SYSCTL_DESCR("Whether the POSIX 1003.1b Memory "
341				    "Protection Option is available on this "
342				    "system"),
343		       NULL, 1, NULL, 0,
344		       CTL_KERN, KERN_MEMORY_PROTECTION, CTL_EOL);
345	sysctl_createv(clog, 0, NULL, NULL,
346		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
347		       CTLTYPE_INT, "login_name_max",
348		       SYSCTL_DESCR("Maximum login name length"),
349		       NULL, LOGIN_NAME_MAX, NULL, 0,
350		       CTL_KERN, KERN_LOGIN_NAME_MAX, CTL_EOL);
351	sysctl_createv(clog, 0, NULL, NULL,
352		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
353		       CTLTYPE_STRING, "defcorename",
354		       SYSCTL_DESCR("Default core file name"),
355		       sysctl_kern_defcorename, 0, defcorename, MAXPATHLEN,
356		       CTL_KERN, KERN_DEFCORENAME, CTL_EOL);
357	sysctl_createv(clog, 0, NULL, NULL,
358		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
359		       CTLTYPE_INT, "logsigexit",
360		       SYSCTL_DESCR("Log process exit when caused by signals"),
361		       NULL, 0, &kern_logsigexit, 0,
362		       CTL_KERN, KERN_LOGSIGEXIT, CTL_EOL);
363	sysctl_createv(clog, 0, NULL, NULL,
364		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
365		       CTLTYPE_INT, "fscale",
366		       SYSCTL_DESCR("Kernel fixed-point scale factor"),
367		       NULL, FSCALE, NULL, 0,
368		       CTL_KERN, KERN_FSCALE, CTL_EOL);
369	sysctl_createv(clog, 0, NULL, NULL,
370		       CTLFLAG_PERMANENT,
371		       CTLTYPE_INT, "ccpu",
372		       SYSCTL_DESCR("Scheduler exponential decay value"),
373		       NULL, 0, &ccpu, 0,
374		       CTL_KERN, KERN_CCPU, CTL_EOL);
375	sysctl_createv(clog, 0, NULL, NULL,
376		       CTLFLAG_PERMANENT,
377		       CTLTYPE_STRUCT, "cp_time",
378		       SYSCTL_DESCR("Clock ticks spent in different CPU states"),
379		       sysctl_kern_cptime, 0, NULL, 0,
380		       CTL_KERN, KERN_CP_TIME, CTL_EOL);
381	sysctl_createv(clog, 0, NULL, NULL,
382		       CTLFLAG_PERMANENT,
383		       CTLTYPE_STRUCT, "consdev",
384		       SYSCTL_DESCR("Console device"),
385		       sysctl_consdev, 0, NULL, sizeof(dev_t),
386		       CTL_KERN, KERN_CONSDEV, CTL_EOL);
387#if NPTY > 0
388	sysctl_createv(clog, 0, NULL, NULL,
389		       CTLFLAG_PERMANENT,
390		       CTLTYPE_INT, "maxptys",
391		       SYSCTL_DESCR("Maximum number of pseudo-ttys"),
392		       sysctl_kern_maxptys, 0, NULL, 0,
393		       CTL_KERN, KERN_MAXPTYS, CTL_EOL);
394#endif /* NPTY > 0 */
395	sysctl_createv(clog, 0, NULL, NULL,
396		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
397		       CTLTYPE_INT, "maxphys",
398		       SYSCTL_DESCR("Maximum raw I/O transfer size"),
399		       NULL, MAXPHYS, NULL, 0,
400		       CTL_KERN, KERN_MAXPHYS, CTL_EOL);
401	sysctl_createv(clog, 0, NULL, NULL,
402		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
403		       CTLTYPE_INT, "monotonic_clock",
404		       SYSCTL_DESCR("Implementation version of the POSIX "
405				    "1003.1b Monotonic Clock Option"),
406		       /* XXX _POSIX_VERSION */
407		       NULL, _POSIX_MONOTONIC_CLOCK, NULL, 0,
408		       CTL_KERN, KERN_MONOTONIC_CLOCK, CTL_EOL);
409	sysctl_createv(clog, 0, NULL, NULL,
410		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
411		       CTLTYPE_INT, "labelsector",
412		       SYSCTL_DESCR("Sector number containing the disklabel"),
413		       NULL, LABELSECTOR, NULL, 0,
414		       CTL_KERN, KERN_LABELSECTOR, CTL_EOL);
415	sysctl_createv(clog, 0, NULL, NULL,
416		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
417		       CTLTYPE_INT, "labeloffset",
418		       SYSCTL_DESCR("Offset of the disklabel within the "
419				    "sector"),
420		       NULL, LABELOFFSET, NULL, 0,
421		       CTL_KERN, KERN_LABELOFFSET, CTL_EOL);
422	sysctl_createv(clog, 0, NULL, NULL,
423		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
424		       CTLTYPE_INT, "labelusesmbr",
425		       SYSCTL_DESCR("disklabel is inside MBR partition"),
426		       NULL, LABELUSESMBR, NULL, 0,
427		       CTL_KERN, CTL_CREATE, CTL_EOL);
428	sysctl_createv(clog, 0, NULL, NULL,
429		       CTLFLAG_PERMANENT,
430		       CTLTYPE_NODE, "lwp",
431		       SYSCTL_DESCR("System-wide LWP information"),
432		       sysctl_kern_lwp, 0, NULL, 0,
433		       CTL_KERN, KERN_LWP, CTL_EOL);
434	sysctl_createv(clog, 0, NULL, NULL,
435		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
436		       CTLTYPE_INT, "forkfsleep",
437		       SYSCTL_DESCR("Milliseconds to sleep on fork failure due "
438				    "to process limits"),
439		       sysctl_kern_forkfsleep, 0, NULL, 0,
440		       CTL_KERN, KERN_FORKFSLEEP, CTL_EOL);
441	sysctl_createv(clog, 0, NULL, NULL,
442		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
443		       CTLTYPE_INT, "posix_threads",
444		       SYSCTL_DESCR("Version of IEEE Std 1003.1 and its "
445				    "Threads option to which the system "
446				    "attempts to conform"),
447		       /* XXX _POSIX_VERSION */
448		       NULL, _POSIX_THREADS, NULL, 0,
449		       CTL_KERN, KERN_POSIX_THREADS, CTL_EOL);
450	sysctl_createv(clog, 0, NULL, NULL,
451		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
452		       CTLTYPE_INT, "posix_semaphores",
453		       SYSCTL_DESCR("Version of IEEE Std 1003.1 and its "
454				    "Semaphores option to which the system "
455				    "attempts to conform"), NULL,
456		       200112, NULL, 0,
457		       CTL_KERN, KERN_POSIX_SEMAPHORES, CTL_EOL);
458	sysctl_createv(clog, 0, NULL, NULL,
459		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
460		       CTLTYPE_INT, "posix_barriers",
461		       SYSCTL_DESCR("Version of IEEE Std 1003.1 and its "
462				    "Barriers option to which the system "
463				    "attempts to conform"),
464		       /* XXX _POSIX_VERSION */
465		       NULL, _POSIX_BARRIERS, NULL, 0,
466		       CTL_KERN, KERN_POSIX_BARRIERS, CTL_EOL);
467	sysctl_createv(clog, 0, NULL, NULL,
468		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
469		       CTLTYPE_INT, "posix_timers",
470		       SYSCTL_DESCR("Version of IEEE Std 1003.1 and its "
471				    "Timers option to which the system "
472				    "attempts to conform"),
473		       /* XXX _POSIX_VERSION */
474		       NULL, _POSIX_TIMERS, NULL, 0,
475		       CTL_KERN, KERN_POSIX_TIMERS, CTL_EOL);
476	sysctl_createv(clog, 0, NULL, NULL,
477		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
478		       CTLTYPE_INT, "posix_spin_locks",
479		       SYSCTL_DESCR("Version of IEEE Std 1003.1 and its Spin "
480				    "Locks option to which the system attempts "
481				    "to conform"),
482		       /* XXX _POSIX_VERSION */
483		       NULL, _POSIX_SPIN_LOCKS, NULL, 0,
484		       CTL_KERN, KERN_POSIX_SPIN_LOCKS, CTL_EOL);
485	sysctl_createv(clog, 0, NULL, NULL,
486		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
487		       CTLTYPE_INT, "posix_reader_writer_locks",
488		       SYSCTL_DESCR("Version of IEEE Std 1003.1 and its "
489				    "Read-Write Locks option to which the "
490				    "system attempts to conform"),
491		       /* XXX _POSIX_VERSION */
492		       NULL, _POSIX_READER_WRITER_LOCKS, NULL, 0,
493		       CTL_KERN, KERN_POSIX_READER_WRITER_LOCKS, CTL_EOL);
494	sysctl_createv(clog, 0, NULL, NULL,
495		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
496		       CTLTYPE_INT, "dump_on_panic",
497		       SYSCTL_DESCR("Perform a crash dump on system panic"),
498		       NULL, 0, &dumponpanic, 0,
499		       CTL_KERN, KERN_DUMP_ON_PANIC, CTL_EOL);
500	sysctl_createv(clog, 0, NULL, NULL,
501		       CTLFLAG_PERMANENT,
502		       CTLTYPE_INT, "root_partition",
503		       SYSCTL_DESCR("Root partition on the root device"),
504		       sysctl_kern_root_partition, 0, NULL, 0,
505		       CTL_KERN, KERN_ROOT_PARTITION, CTL_EOL);
506	sysctl_createv(clog, 0, NULL, NULL,
507		       CTLFLAG_PERMANENT,
508		       CTLTYPE_STRUCT, "drivers",
509		       SYSCTL_DESCR("List of all drivers with block and "
510				    "character device numbers"),
511		       sysctl_kern_drivers, 0, NULL, 0,
512		       CTL_KERN, KERN_DRIVERS, CTL_EOL);
513	sysctl_createv(clog, 0, NULL, NULL,
514		       CTLFLAG_PERMANENT,
515		       CTLTYPE_STRUCT, "cp_id",
516		       SYSCTL_DESCR("Mapping of CPU number to CPU id"),
517		       sysctl_kern_cpid, 0, NULL, 0,
518		       CTL_KERN, KERN_CP_ID, CTL_EOL);
519	sysctl_createv(clog, 0, NULL, &rnode,
520		       CTLFLAG_PERMANENT,
521		       CTLTYPE_NODE, "coredump",
522		       SYSCTL_DESCR("Coredump settings."),
523		       NULL, 0, NULL, 0,
524		       CTL_KERN, CTL_CREATE, CTL_EOL);
525	sysctl_createv(clog, 0, &rnode, &rnode,
526		       CTLFLAG_PERMANENT,
527		       CTLTYPE_NODE, "setid",
528		       SYSCTL_DESCR("Set-id processes' coredump settings."),
529		       NULL, 0, NULL, 0,
530		       CTL_CREATE, CTL_EOL);
531	sysctl_createv(clog, 0, &rnode, NULL,
532		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
533		       CTLTYPE_INT, "dump",
534		       SYSCTL_DESCR("Allow set-id processes to dump core."),
535		       sysctl_security_setidcore, 0, &security_setidcore_dump,
536		       sizeof(security_setidcore_dump),
537		       CTL_CREATE, CTL_EOL);
538	sysctl_createv(clog, 0, &rnode, NULL,
539		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
540		       CTLTYPE_STRING, "path",
541		       SYSCTL_DESCR("Path pattern for set-id coredumps."),
542		       sysctl_security_setidcorename, 0,
543		       security_setidcore_path,
544		       sizeof(security_setidcore_path),
545		       CTL_CREATE, CTL_EOL);
546	sysctl_createv(clog, 0, &rnode, NULL,
547		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
548		       CTLTYPE_INT, "owner",
549		       SYSCTL_DESCR("Owner id for set-id processes' cores."),
550		       sysctl_security_setidcore, 0, &security_setidcore_owner,
551		       0,
552		       CTL_CREATE, CTL_EOL);
553	sysctl_createv(clog, 0, &rnode, NULL,
554		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
555		       CTLTYPE_INT, "group",
556		       SYSCTL_DESCR("Group id for set-id processes' cores."),
557		       sysctl_security_setidcore, 0, &security_setidcore_group,
558		       0,
559		       CTL_CREATE, CTL_EOL);
560	sysctl_createv(clog, 0, &rnode, NULL,
561		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
562		       CTLTYPE_INT, "mode",
563		       SYSCTL_DESCR("Mode for set-id processes' cores."),
564		       sysctl_security_setidcore, 0, &security_setidcore_mode,
565		       0,
566		       CTL_CREATE, CTL_EOL);
567	sysctl_createv(clog, 0, NULL, NULL,
568		       CTLFLAG_IMMEDIATE|CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
569		       CTLTYPE_INT, "no_sa_support",
570		       SYSCTL_DESCR("0 if the kernel supports SA, otherwise "
571		       "it doesn't"),
572		       NULL, 1, NULL, 0,
573		       CTL_KERN, CTL_CREATE, CTL_EOL);
574	sysctl_createv(clog, 0, NULL, NULL,
575			CTLFLAG_PERMANENT,
576			CTLTYPE_STRING, "configname",
577			SYSCTL_DESCR("Name of config file"),
578			NULL, 0, __UNCONST(kernel_ident), 0,
579			CTL_KERN, CTL_CREATE, CTL_EOL);
580	sysctl_createv(clog, 0, NULL, NULL,
581			CTLFLAG_PERMANENT,
582			CTLTYPE_STRING, "buildinfo",
583			SYSCTL_DESCR("Information from build environment"),
584			NULL, 0, __UNCONST(buildinfo), 0,
585			CTL_KERN, CTL_CREATE, CTL_EOL);
586	sysctl_createv(clog, 0, NULL, NULL,
587			CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
588			CTLTYPE_INT, "messages",
589			SYSCTL_DESCR("Kernel message verbosity"),
590			sysctl_kern_messages, 0, NULL, 0,
591			CTL_KERN, CTL_CREATE, CTL_EOL);
592}
593
594SYSCTL_SETUP(sysctl_hw_misc_setup, "sysctl hw subtree misc setup")
595{
596
597	sysctl_createv(clog, 0, NULL, NULL,
598		       CTLFLAG_PERMANENT,
599		       CTLTYPE_INT, "usermem",
600		       SYSCTL_DESCR("Bytes of non-kernel memory"),
601		       sysctl_hw_usermem, 0, NULL, 0,
602		       CTL_HW, HW_USERMEM, CTL_EOL);
603	sysctl_createv(clog, 0, NULL, NULL,
604		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE|CTLFLAG_HEX,
605		       CTLTYPE_STRING, "cnmagic",
606		       SYSCTL_DESCR("Console magic key sequence"),
607		       sysctl_hw_cnmagic, 0, NULL, CNS_LEN,
608		       CTL_HW, HW_CNMAGIC, CTL_EOL);
609	sysctl_createv(clog, 0, NULL, NULL,
610		       CTLFLAG_PERMANENT,
611		       CTLTYPE_QUAD, "usermem64",
612		       SYSCTL_DESCR("Bytes of non-kernel memory"),
613		       sysctl_hw_usermem, 0, NULL, 0,
614		       CTL_HW, HW_USERMEM64, CTL_EOL);
615}
616
617#ifdef DEBUG
618/*
619 * Debugging related system variables.
620 */
621struct ctldebug /* debug0, */ /* debug1, */ debug2, debug3, debug4;
622struct ctldebug debug5, debug6, debug7, debug8, debug9;
623struct ctldebug debug10, debug11, debug12, debug13, debug14;
624struct ctldebug debug15, debug16, debug17, debug18, debug19;
625static struct ctldebug *debugvars[] = {
626	&debug0, &debug1, &debug2, &debug3, &debug4,
627	&debug5, &debug6, &debug7, &debug8, &debug9,
628	&debug10, &debug11, &debug12, &debug13, &debug14,
629	&debug15, &debug16, &debug17, &debug18, &debug19,
630};
631
632/*
633 * this setup routine is a replacement for debug_sysctl()
634 *
635 * note that it creates several nodes per defined debug variable
636 */
637SYSCTL_SETUP(sysctl_debug_setup, "sysctl debug subtree setup")
638{
639	struct ctldebug *cdp;
640	char nodename[20];
641	int i;
642
643	/*
644	 * two ways here:
645	 *
646	 * the "old" way (debug.name -> value) which was emulated by
647	 * the sysctl(8) binary
648	 *
649	 * the new way, which the sysctl(8) binary was actually using
650
651	 node	debug
652	 node	debug.0
653	 string debug.0.name
654	 int	debug.0.value
655	 int	debug.name
656
657	 */
658
659	for (i = 0; i < __arraycount(debugvars); i++) {
660		cdp = debugvars[i];
661		if (cdp->debugname == NULL || cdp->debugvar == NULL)
662			continue;
663
664		snprintf(nodename, sizeof(nodename), "debug%d", i);
665		sysctl_createv(clog, 0, NULL, NULL,
666			       CTLFLAG_PERMANENT|CTLFLAG_HIDDEN,
667			       CTLTYPE_NODE, nodename, NULL,
668			       NULL, 0, NULL, 0,
669			       CTL_DEBUG, i, CTL_EOL);
670		sysctl_createv(clog, 0, NULL, NULL,
671			       CTLFLAG_PERMANENT|CTLFLAG_HIDDEN,
672			       CTLTYPE_STRING, "name", NULL,
673			       /*XXXUNCONST*/
674			       NULL, 0, __UNCONST(cdp->debugname), 0,
675			       CTL_DEBUG, i, CTL_DEBUG_NAME, CTL_EOL);
676		sysctl_createv(clog, 0, NULL, NULL,
677			       CTLFLAG_PERMANENT|CTLFLAG_HIDDEN,
678			       CTLTYPE_INT, "value", NULL,
679			       NULL, 0, cdp->debugvar, 0,
680			       CTL_DEBUG, i, CTL_DEBUG_VALUE, CTL_EOL);
681		sysctl_createv(clog, 0, NULL, NULL,
682			       CTLFLAG_PERMANENT,
683			       CTLTYPE_INT, cdp->debugname, NULL,
684			       NULL, 0, cdp->debugvar, 0,
685			       CTL_DEBUG, CTL_CREATE, CTL_EOL);
686	}
687}
688#endif /* DEBUG */
689
690/*
691 * ********************************************************************
692 * section 2: private node-specific helper routines.
693 * ********************************************************************
694 */
695
696/*
697 * sysctl helper routine for kern.maxvnodes.  Drain vnodes if
698 * new value is lower than desiredvnodes and then calls reinit
699 * routines that needs to adjust to the new value.
700 */
701static int
702sysctl_kern_maxvnodes(SYSCTLFN_ARGS)
703{
704	int error, new_vnodes, old_vnodes, new_max;
705	struct sysctlnode node;
706
707	new_vnodes = desiredvnodes;
708	node = *rnode;
709	node.sysctl_data = &new_vnodes;
710	error = sysctl_lookup(SYSCTLFN_CALL(&node));
711	if (error || newp == NULL)
712		return (error);
713
714	/*
715	 * sysctl passes down unsigned values, require them
716	 * to be positive
717	 */
718	if (new_vnodes <= 0)
719		return (EINVAL);
720
721	/* Limits: 75% of kmem and physical memory. */
722	new_max = calc_cache_size(vmem_size(kmem_arena, VMEM_FREE|VMEM_ALLOC),
723	    75, 75) / VNODE_COST;
724	if (new_vnodes > new_max)
725		new_vnodes = new_max;
726
727	old_vnodes = desiredvnodes;
728	desiredvnodes = new_vnodes;
729	error = vfs_drainvnodes();
730	if (error) {
731		desiredvnodes = old_vnodes;
732		return (error);
733	}
734	vfs_reinit();
735	nchreinit();
736
737	return (0);
738}
739
740/*
741 * sysctl helper routine for kern.messages.
742 * Alters boothowto to display kernel messages in increasing verbosity
743 * from 0 to 4.
744 */
745
746#define MAXMESSAGES            4
747static int
748sysctl_kern_messages(SYSCTLFN_ARGS)
749{
750	int error, messageverbose, messagemask, newboothowto;
751	struct sysctlnode node;
752
753	messagemask = (AB_NORMAL|AB_QUIET|AB_SILENT|AB_VERBOSE|AB_DEBUG);
754	switch (boothowto & messagemask) {
755	case AB_SILENT:
756		messageverbose = 0;
757		break;
758	case AB_QUIET:
759		messageverbose = 1;
760		break;
761	case AB_VERBOSE:
762		messageverbose = 3;
763		break;
764	case AB_DEBUG:
765		messageverbose = 4;
766		break;
767	case AB_NORMAL:
768	default:
769		messageverbose = 2;
770	}
771
772	node = *rnode;
773	node.sysctl_data = &messageverbose;
774	error = sysctl_lookup(SYSCTLFN_CALL(&node));
775	if (error || newp == NULL)
776		return (error);
777	if (messageverbose < 0 || messageverbose > MAXMESSAGES)
778		return EINVAL;
779
780	/* Set boothowto */
781	newboothowto = boothowto & ~messagemask;
782
783	switch (messageverbose) {
784	case 0:
785		newboothowto |= AB_SILENT;
786		break;
787	case 1:
788		newboothowto |= AB_QUIET;
789		break;
790	case 3:
791		newboothowto |= AB_VERBOSE;
792		break;
793	case 4:
794		newboothowto |= AB_DEBUG;
795		break;
796	case 2:
797	default:                /* Messages default to normal. */
798		break;
799	}
800
801	boothowto = newboothowto;
802
803	return (0);
804}
805
806/*
807 * sysctl helper routine for the kern.boottime node
808 */
809static int
810sysctl_kern_boottime(SYSCTLFN_ARGS)
811{
812	struct sysctlnode node;
813	struct timespec ts;
814
815	getnanoboottime(&ts);
816	node = *rnode;
817	node.sysctl_data = &ts;
818	return (sysctl_lookup(SYSCTLFN_CALL(&node)));
819}
820
821/*
822 * sysctl helper routine for rtc_offset - set time after changes
823 */
824static int
825sysctl_kern_rtc_offset(SYSCTLFN_ARGS)
826{
827	struct timespec ts, delta;
828	int error, new_rtc_offset;
829	struct sysctlnode node;
830
831	new_rtc_offset = rtc_offset;
832	node = *rnode;
833	node.sysctl_data = &new_rtc_offset;
834	error = sysctl_lookup(SYSCTLFN_CALL(&node));
835	if (error || newp == NULL)
836		return (error);
837
838	if (kauth_authorize_system(l->l_cred, KAUTH_SYSTEM_TIME,
839	    KAUTH_REQ_SYSTEM_TIME_RTCOFFSET,
840	    KAUTH_ARG(new_rtc_offset), NULL, NULL))
841		return (EPERM);
842	if (rtc_offset == new_rtc_offset)
843		return (0);
844
845	/* if we change the offset, adjust the time */
846	nanotime(&ts);
847	delta.tv_sec = 60 * (new_rtc_offset - rtc_offset);
848	delta.tv_nsec = 0;
849	timespecadd(&ts, &delta, &ts);
850	rtc_offset = new_rtc_offset;
851	return (settime(l->l_proc, &ts));
852}
853
854/*
855 * sysctl helper routine for kern.maxproc. Ensures that the new
856 * values are not too low or too high.
857 */
858static int
859sysctl_kern_maxproc(SYSCTLFN_ARGS)
860{
861	int error, nmaxproc;
862	struct sysctlnode node;
863
864	nmaxproc = maxproc;
865	node = *rnode;
866	node.sysctl_data = &nmaxproc;
867	error = sysctl_lookup(SYSCTLFN_CALL(&node));
868	if (error || newp == NULL)
869		return (error);
870
871	if (nmaxproc < 0 || nmaxproc >= PID_MAX)
872		return (EINVAL);
873#ifdef __HAVE_CPU_MAXPROC
874	if (nmaxproc > cpu_maxproc())
875		return (EINVAL);
876#endif
877	error = 0;
878#ifdef __HAVE_MAXPROC_HOOK
879	error = cpu_maxproc_hook(nmaxproc);
880#endif
881	if (error)
882		return error;
883
884	maxproc = nmaxproc;
885
886	return (0);
887}
888
889/*
890 * sysctl helper function for kern.hostid. The hostid is a long, but
891 * we export it as an int, so we need to give it a little help.
892 */
893static int
894sysctl_kern_hostid(SYSCTLFN_ARGS)
895{
896	int error, inthostid;
897	struct sysctlnode node;
898
899	inthostid = hostid;  /* XXX assumes sizeof int <= sizeof long */
900	node = *rnode;
901	node.sysctl_data = &inthostid;
902	error = sysctl_lookup(SYSCTLFN_CALL(&node));
903	if (error || newp == NULL)
904		return (error);
905
906	hostid = (unsigned)inthostid;
907
908	return (0);
909}
910
911/*
912 * sysctl helper routine for kern.defcorename. In the case of a new
913 * string being assigned, check that it's not a zero-length string.
914 * (XXX the check in -current doesn't work, but do we really care?)
915 */
916static int
917sysctl_kern_defcorename(SYSCTLFN_ARGS)
918{
919	int error;
920	char *newcorename;
921	struct sysctlnode node;
922
923	newcorename = PNBUF_GET();
924	node = *rnode;
925	node.sysctl_data = &newcorename[0];
926	memcpy(node.sysctl_data, rnode->sysctl_data, MAXPATHLEN);
927	error = sysctl_lookup(SYSCTLFN_CALL(&node));
928	if (error || newp == NULL) {
929		goto done;
930	}
931
932	/*
933	 * when sysctl_lookup() deals with a string, it's guaranteed
934	 * to come back nul terminated. So there.  :)
935	 */
936	if (strlen(newcorename) == 0) {
937		error = EINVAL;
938	} else {
939		memcpy(rnode->sysctl_data, node.sysctl_data, MAXPATHLEN);
940		error = 0;
941	}
942done:
943	PNBUF_PUT(newcorename);
944	return error;
945}
946
947/*
948 * sysctl helper routine for kern.cp_time node. Adds up cpu time
949 * across all cpus.
950 */
951static int
952sysctl_kern_cptime(SYSCTLFN_ARGS)
953{
954	struct sysctlnode node = *rnode;
955	uint64_t *cp_time = NULL;
956	int error, n = ncpu, i;
957	struct cpu_info *ci;
958	CPU_INFO_ITERATOR cii;
959
960	/*
961	 * if you specifically pass a buffer that is the size of the
962	 * sum, or if you are probing for the size, you get the "sum"
963	 * of cp_time (and the size thereof) across all processors.
964	 *
965	 * alternately, you can pass an additional mib number and get
966	 * cp_time for that particular processor.
967	 */
968	switch (namelen) {
969	case 0:
970		if (*oldlenp == sizeof(uint64_t) * CPUSTATES || oldp == NULL) {
971			node.sysctl_size = sizeof(uint64_t) * CPUSTATES;
972			n = -1; /* SUM */
973		}
974		else {
975			node.sysctl_size = n * sizeof(uint64_t) * CPUSTATES;
976			n = -2; /* ALL */
977		}
978		break;
979	case 1:
980		if (name[0] < 0 || name[0] >= n)
981			return (ENOENT); /* ENOSUCHPROCESSOR */
982		node.sysctl_size = sizeof(uint64_t) * CPUSTATES;
983		n = name[0];
984		/*
985		 * adjust these so that sysctl_lookup() will be happy
986		 */
987		name++;
988		namelen--;
989		break;
990	default:
991		return (EINVAL);
992	}
993
994	cp_time = kmem_alloc(node.sysctl_size, KM_SLEEP);
995	node.sysctl_data = cp_time;
996	memset(cp_time, 0, node.sysctl_size);
997
998	for (CPU_INFO_FOREACH(cii, ci)) {
999		if (n <= 0) {
1000			for (i = 0; i < CPUSTATES; i++) {
1001				cp_time[i] += ci->ci_schedstate.spc_cp_time[i];
1002			}
1003		}
1004		/*
1005		 * if a specific processor was requested and we just
1006		 * did it, we're done here
1007		 */
1008		if (n == 0)
1009			break;
1010		/*
1011		 * if doing "all", skip to next cp_time set for next processor
1012		 */
1013		if (n == -2)
1014			cp_time += CPUSTATES;
1015		/*
1016		 * if we're doing a specific processor, we're one
1017		 * processor closer
1018		 */
1019		if (n > 0)
1020			n--;
1021	}
1022
1023	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1024	kmem_free(node.sysctl_data, node.sysctl_size);
1025	return (error);
1026}
1027
1028#if NPTY > 0
1029/*
1030 * sysctl helper routine for kern.maxptys. Ensures that any new value
1031 * is acceptable to the pty subsystem.
1032 */
1033static int
1034sysctl_kern_maxptys(SYSCTLFN_ARGS)
1035{
1036	int pty_maxptys(int, int);		/* defined in kern/tty_pty.c */
1037	int error, xmax;
1038	struct sysctlnode node;
1039
1040	/* get current value of maxptys */
1041	xmax = pty_maxptys(0, 0);
1042
1043	node = *rnode;
1044	node.sysctl_data = &xmax;
1045	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1046	if (error || newp == NULL)
1047		return (error);
1048
1049	if (xmax != pty_maxptys(xmax, 1))
1050		return (EINVAL);
1051
1052	return (0);
1053}
1054#endif /* NPTY > 0 */
1055
1056/*
1057 * sysctl helper routine to do kern.lwp.* work.
1058 */
1059static int
1060sysctl_kern_lwp(SYSCTLFN_ARGS)
1061{
1062	struct kinfo_lwp klwp;
1063	struct proc *p;
1064	struct lwp *l2, *l3;
1065	char *where, *dp;
1066	int pid, elem_size, elem_count;
1067	int buflen, needed, error;
1068	bool gotit;
1069
1070	if (namelen == 1 && name[0] == CTL_QUERY)
1071		return (sysctl_query(SYSCTLFN_CALL(rnode)));
1072
1073	dp = where = oldp;
1074	buflen = where != NULL ? *oldlenp : 0;
1075	error = needed = 0;
1076
1077	if (newp != NULL || namelen != 3)
1078		return (EINVAL);
1079	pid = name[0];
1080	elem_size = name[1];
1081	elem_count = name[2];
1082
1083	sysctl_unlock();
1084	if (pid == -1) {
1085		mutex_enter(proc_lock);
1086		PROCLIST_FOREACH(p, &allproc) {
1087			/* Grab a hold on the process. */
1088			if (!rw_tryenter(&p->p_reflock, RW_READER)) {
1089				continue;
1090			}
1091			mutex_exit(proc_lock);
1092
1093			mutex_enter(p->p_lock);
1094			LIST_FOREACH(l2, &p->p_lwps, l_sibling) {
1095				if (buflen >= elem_size && elem_count > 0) {
1096					lwp_lock(l2);
1097					fill_lwp(l2, &klwp);
1098					lwp_unlock(l2);
1099					mutex_exit(p->p_lock);
1100
1101					/*
1102					 * Copy out elem_size, but not
1103					 * larger than the size of a
1104					 * struct kinfo_proc2.
1105					 */
1106					error = dcopyout(l, &klwp, dp,
1107					    uimin(sizeof(klwp), elem_size));
1108					if (error) {
1109						rw_exit(&p->p_reflock);
1110						goto cleanup;
1111					}
1112					mutex_enter(p->p_lock);
1113					LIST_FOREACH(l3, &p->p_lwps,
1114					    l_sibling) {
1115						if (l2 == l3)
1116							break;
1117					}
1118					if (l3 == NULL) {
1119						mutex_exit(p->p_lock);
1120						rw_exit(&p->p_reflock);
1121						error = EAGAIN;
1122						goto cleanup;
1123					}
1124					dp += elem_size;
1125					buflen -= elem_size;
1126					elem_count--;
1127				}
1128				needed += elem_size;
1129			}
1130			mutex_exit(p->p_lock);
1131
1132			/* Drop reference to process. */
1133			mutex_enter(proc_lock);
1134			rw_exit(&p->p_reflock);
1135		}
1136		mutex_exit(proc_lock);
1137	} else {
1138		mutex_enter(proc_lock);
1139		p = proc_find(pid);
1140		if (p == NULL) {
1141			error = ESRCH;
1142			mutex_exit(proc_lock);
1143			goto cleanup;
1144		}
1145		/* Grab a hold on the process. */
1146		gotit = rw_tryenter(&p->p_reflock, RW_READER);
1147		mutex_exit(proc_lock);
1148		if (!gotit) {
1149			error = ESRCH;
1150			goto cleanup;
1151		}
1152
1153		mutex_enter(p->p_lock);
1154		LIST_FOREACH(l2, &p->p_lwps, l_sibling) {
1155			if (buflen >= elem_size && elem_count > 0) {
1156				lwp_lock(l2);
1157				fill_lwp(l2, &klwp);
1158				lwp_unlock(l2);
1159				mutex_exit(p->p_lock);
1160				/*
1161				 * Copy out elem_size, but not larger than
1162				 * the size of a struct kinfo_proc2.
1163				 */
1164				error = dcopyout(l, &klwp, dp,
1165				    uimin(sizeof(klwp), elem_size));
1166				if (error) {
1167					rw_exit(&p->p_reflock);
1168					goto cleanup;
1169				}
1170				mutex_enter(p->p_lock);
1171				LIST_FOREACH(l3, &p->p_lwps, l_sibling) {
1172					if (l2 == l3)
1173						break;
1174				}
1175				if (l3 == NULL) {
1176					mutex_exit(p->p_lock);
1177					rw_exit(&p->p_reflock);
1178					error = EAGAIN;
1179					goto cleanup;
1180				}
1181				dp += elem_size;
1182				buflen -= elem_size;
1183				elem_count--;
1184			}
1185			needed += elem_size;
1186		}
1187		mutex_exit(p->p_lock);
1188
1189		/* Drop reference to process. */
1190		rw_exit(&p->p_reflock);
1191	}
1192
1193	if (where != NULL) {
1194		*oldlenp = dp - where;
1195		if (needed > *oldlenp) {
1196			sysctl_relock();
1197			return (ENOMEM);
1198		}
1199	} else {
1200		needed += KERN_LWPSLOP;
1201		*oldlenp = needed;
1202	}
1203	error = 0;
1204 cleanup:
1205	sysctl_relock();
1206	return (error);
1207}
1208
1209/*
1210 * sysctl helper routine for kern.forkfsleep node. Ensures that the
1211 * given value is not too large or two small, and is at least one
1212 * timer tick if not zero.
1213 */
1214static int
1215sysctl_kern_forkfsleep(SYSCTLFN_ARGS)
1216{
1217	/* userland sees value in ms, internally is in ticks */
1218	extern int forkfsleep;		/* defined in kern/kern_fork.c */
1219	int error, timo, lsleep;
1220	struct sysctlnode node;
1221
1222	lsleep = forkfsleep * 1000 / hz;
1223	node = *rnode;
1224	node.sysctl_data = &lsleep;
1225	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1226	if (error || newp == NULL)
1227		return (error);
1228
1229	/* refuse negative values, and overly 'long time' */
1230	if (lsleep < 0 || lsleep > MAXSLP * 1000)
1231		return (EINVAL);
1232
1233	timo = mstohz(lsleep);
1234
1235	/* if the interval is >0 ms && <1 tick, use 1 tick */
1236	if (lsleep != 0 && timo == 0)
1237		forkfsleep = 1;
1238	else
1239		forkfsleep = timo;
1240
1241	return (0);
1242}
1243
1244/*
1245 * sysctl helper routine for kern.root_partition
1246 */
1247static int
1248sysctl_kern_root_partition(SYSCTLFN_ARGS)
1249{
1250	int rootpart = DISKPART(rootdev);
1251	struct sysctlnode node = *rnode;
1252
1253	node.sysctl_data = &rootpart;
1254	return (sysctl_lookup(SYSCTLFN_CALL(&node)));
1255}
1256
1257/*
1258 * sysctl helper function for kern.drivers
1259 */
1260static int
1261sysctl_kern_drivers(SYSCTLFN_ARGS)
1262{
1263	int error;
1264	size_t buflen;
1265	struct kinfo_drivers kd;
1266	char *start, *where;
1267	const char *dname;
1268	int i;
1269	extern struct devsw_conv *devsw_conv;
1270	extern int max_devsw_convs;
1271
1272	start = where = oldp;
1273	buflen = *oldlenp;
1274	if (where == NULL) {
1275		*oldlenp = max_devsw_convs * sizeof kd;
1276		return 0;
1277	}
1278
1279	/*
1280	 * An array of kinfo_drivers structures
1281	 */
1282	error = 0;
1283	sysctl_unlock();
1284	mutex_enter(&device_lock);
1285	for (i = 0; i < max_devsw_convs; i++) {
1286		dname = devsw_conv[i].d_name;
1287		if (dname == NULL)
1288			continue;
1289		if (buflen < sizeof kd) {
1290			error = ENOMEM;
1291			break;
1292		}
1293		memset(&kd, 0, sizeof(kd));
1294		kd.d_bmajor = devsw_conv[i].d_bmajor;
1295		kd.d_cmajor = devsw_conv[i].d_cmajor;
1296		strlcpy(kd.d_name, dname, sizeof kd.d_name);
1297		mutex_exit(&device_lock);
1298		error = dcopyout(l, &kd, where, sizeof kd);
1299		mutex_enter(&device_lock);
1300		if (error != 0)
1301			break;
1302		buflen -= sizeof kd;
1303		where += sizeof kd;
1304	}
1305	mutex_exit(&device_lock);
1306	sysctl_relock();
1307	*oldlenp = where - start;
1308	return error;
1309}
1310
1311static int
1312sysctl_security_setidcore(SYSCTLFN_ARGS)
1313{
1314	int newsize, error;
1315	struct sysctlnode node;
1316
1317	node = *rnode;
1318	node.sysctl_data = &newsize;
1319	newsize = *(int *)rnode->sysctl_data;
1320	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1321	if (error || newp == NULL)
1322		return error;
1323
1324	if (kauth_authorize_system(l->l_cred, KAUTH_SYSTEM_SETIDCORE,
1325	    0, NULL, NULL, NULL))
1326		return (EPERM);
1327
1328	*(int *)rnode->sysctl_data = newsize;
1329
1330	return 0;
1331}
1332
1333static int
1334sysctl_security_setidcorename(SYSCTLFN_ARGS)
1335{
1336	int error;
1337	char *newsetidcorename;
1338	struct sysctlnode node;
1339
1340	newsetidcorename = PNBUF_GET();
1341	node = *rnode;
1342	node.sysctl_data = newsetidcorename;
1343	memcpy(node.sysctl_data, rnode->sysctl_data, MAXPATHLEN);
1344	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1345	if (error || newp == NULL) {
1346		goto out;
1347	}
1348	if (kauth_authorize_system(l->l_cred, KAUTH_SYSTEM_SETIDCORE,
1349	    0, NULL, NULL, NULL)) {
1350		error = EPERM;
1351		goto out;
1352	}
1353	if (strlen(newsetidcorename) == 0) {
1354		error = EINVAL;
1355		goto out;
1356	}
1357	memcpy(rnode->sysctl_data, node.sysctl_data, MAXPATHLEN);
1358out:
1359	PNBUF_PUT(newsetidcorename);
1360	return error;
1361}
1362
1363/*
1364 * sysctl helper routine for kern.cp_id node. Maps cpus to their
1365 * cpuids.
1366 */
1367static int
1368sysctl_kern_cpid(SYSCTLFN_ARGS)
1369{
1370	struct sysctlnode node = *rnode;
1371	uint64_t *cp_id = NULL;
1372	int error, n = ncpu;
1373	struct cpu_info *ci;
1374	CPU_INFO_ITERATOR cii;
1375
1376	/*
1377	 * Here you may either retrieve a single cpu id or the whole
1378	 * set. The size you get back when probing depends on what
1379	 * you ask for.
1380	 */
1381	switch (namelen) {
1382	case 0:
1383		node.sysctl_size = n * sizeof(uint64_t);
1384		n = -2; /* ALL */
1385		break;
1386	case 1:
1387		if (name[0] < 0 || name[0] >= n)
1388			return (ENOENT); /* ENOSUCHPROCESSOR */
1389		node.sysctl_size = sizeof(uint64_t);
1390		n = name[0];
1391		/*
1392		 * adjust these so that sysctl_lookup() will be happy
1393		 */
1394		name++;
1395		namelen--;
1396		break;
1397	default:
1398		return (EINVAL);
1399	}
1400
1401	cp_id = kmem_alloc(node.sysctl_size, KM_SLEEP);
1402	node.sysctl_data = cp_id;
1403	memset(cp_id, 0, node.sysctl_size);
1404
1405	for (CPU_INFO_FOREACH(cii, ci)) {
1406		if (n <= 0)
1407			cp_id[0] = cpu_index(ci);
1408		/*
1409		 * if a specific processor was requested and we just
1410		 * did it, we're done here
1411		 */
1412		if (n == 0)
1413			break;
1414		/*
1415		 * if doing "all", skip to next cp_id slot for next processor
1416		 */
1417		if (n == -2)
1418			cp_id++;
1419		/*
1420		 * if we're doing a specific processor, we're one
1421		 * processor closer
1422		 */
1423		if (n > 0)
1424			n--;
1425	}
1426
1427	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1428	kmem_free(node.sysctl_data, node.sysctl_size);
1429	return (error);
1430}
1431
1432/*
1433 * sysctl helper routine for hw.usermem and hw.usermem64. Values are
1434 * calculate on the fly taking into account integer overflow and the
1435 * current wired count.
1436 */
1437static int
1438sysctl_hw_usermem(SYSCTLFN_ARGS)
1439{
1440	u_int ui;
1441	u_quad_t uq;
1442	struct sysctlnode node;
1443
1444	node = *rnode;
1445	switch (rnode->sysctl_num) {
1446	case HW_USERMEM:
1447		if ((ui = physmem - uvmexp.wired) > (UINT_MAX / PAGE_SIZE))
1448			ui = UINT_MAX;
1449		else
1450			ui *= PAGE_SIZE;
1451		node.sysctl_data = &ui;
1452		break;
1453	case HW_USERMEM64:
1454		uq = (u_quad_t)(physmem - uvmexp.wired) * PAGE_SIZE;
1455		node.sysctl_data = &uq;
1456		break;
1457	default:
1458		return (EINVAL);
1459	}
1460
1461	return (sysctl_lookup(SYSCTLFN_CALL(&node)));
1462}
1463
1464/*
1465 * sysctl helper routine for kern.cnmagic node. Pulls the old value
1466 * out, encoded, and stuffs the new value in for decoding.
1467 */
1468static int
1469sysctl_hw_cnmagic(SYSCTLFN_ARGS)
1470{
1471	char magic[CNS_LEN];
1472	int error;
1473	struct sysctlnode node;
1474
1475	if (oldp)
1476		cn_get_magic(magic, CNS_LEN);
1477	node = *rnode;
1478	node.sysctl_data = &magic[0];
1479	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1480	if (error || newp == NULL)
1481		return (error);
1482
1483	return (cn_set_magic(magic));
1484}
1485
1486/*
1487 * ********************************************************************
1488 * section 3: public helper routines that are used for more than one
1489 * node
1490 * ********************************************************************
1491 */
1492
1493/*
1494 * sysctl helper routine for the kern.root_device node and some ports'
1495 * machdep.root_device nodes.
1496 */
1497int
1498sysctl_root_device(SYSCTLFN_ARGS)
1499{
1500	struct sysctlnode node;
1501
1502	node = *rnode;
1503	node.sysctl_data = __UNCONST(device_xname(root_device));
1504	node.sysctl_size = strlen(device_xname(root_device)) + 1;
1505	return (sysctl_lookup(SYSCTLFN_CALL(&node)));
1506}
1507
1508/*
1509 * sysctl helper routine for kern.consdev, dependent on the current
1510 * state of the console. Also used for machdep.console_device on some
1511 * ports.
1512 */
1513int
1514sysctl_consdev(SYSCTLFN_ARGS)
1515{
1516	dev_t consdev;
1517	uint32_t oconsdev;
1518	struct sysctlnode node;
1519
1520	if (cn_tab != NULL)
1521		consdev = cn_tab->cn_dev;
1522	else
1523		consdev = NODEV;
1524	node = *rnode;
1525	switch (*oldlenp) {
1526	case sizeof(consdev):
1527		node.sysctl_data = &consdev;
1528		node.sysctl_size = sizeof(consdev);
1529		break;
1530	case sizeof(oconsdev):
1531		oconsdev = (uint32_t)consdev;
1532		node.sysctl_data = &oconsdev;
1533		node.sysctl_size = sizeof(oconsdev);
1534		break;
1535	default:
1536		return EINVAL;
1537	}
1538	return (sysctl_lookup(SYSCTLFN_CALL(&node)));
1539}
1540
1541/*
1542 * ********************************************************************
1543 * section 4: support for some helpers
1544 * ********************************************************************
1545 */
1546
1547
1548/*
1549 * Fill in a kinfo_lwp structure for the specified lwp.
1550 */
1551static void
1552fill_lwp(struct lwp *l, struct kinfo_lwp *kl)
1553{
1554	const bool allowaddr = get_expose_address(curproc);
1555	struct proc *p = l->l_proc;
1556	struct timeval tv;
1557
1558	KASSERT(lwp_locked(l, NULL));
1559
1560	memset(kl, 0, sizeof(*kl));
1561
1562	kl->l_forw = 0;
1563	kl->l_back = 0;
1564	COND_SET_VALUE(kl->l_laddr, PTRTOUINT64(l), allowaddr);
1565	COND_SET_VALUE(kl->l_addr, PTRTOUINT64(l->l_addr), allowaddr);
1566	kl->l_stat = l->l_stat;
1567	kl->l_lid = l->l_lid;
1568	kl->l_flag = L_INMEM;
1569	kl->l_flag |= sysctl_map_flags(sysctl_lwpprflagmap, l->l_prflag);
1570	kl->l_flag |= sysctl_map_flags(sysctl_lwpflagmap, l->l_flag);
1571
1572	kl->l_swtime = l->l_swtime;
1573	kl->l_slptime = l->l_slptime;
1574	if (l->l_stat == LSONPROC)
1575		kl->l_schedflags = l->l_cpu->ci_schedstate.spc_flags;
1576	else
1577		kl->l_schedflags = 0;
1578	kl->l_priority = lwp_eprio(l);
1579	kl->l_usrpri = l->l_priority;
1580	if (l->l_wchan)
1581		strncpy(kl->l_wmesg, l->l_wmesg, sizeof(kl->l_wmesg));
1582	COND_SET_VALUE(kl->l_wchan, PTRTOUINT64(l->l_wchan), allowaddr);
1583	kl->l_cpuid = cpu_index(l->l_cpu);
1584	bintime2timeval(&l->l_rtime, &tv);
1585	kl->l_rtime_sec = tv.tv_sec;
1586	kl->l_rtime_usec = tv.tv_usec;
1587	kl->l_cpticks = l->l_cpticks;
1588	kl->l_pctcpu = l->l_pctcpu;
1589	kl->l_pid = p->p_pid;
1590	if (l->l_name == NULL)
1591		kl->l_name[0] = '\0';
1592	else
1593		strlcpy(kl->l_name, l->l_name, sizeof(kl->l_name));
1594}
1595