Home | History | Annotate | Line # | Download | only in kern
init_sysctl.c revision 1.129
      1 /*	$NetBSD: init_sysctl.c,v 1.129 2008/04/02 10:53:23 xtraeme 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  * 3. All advertising materials mentioning features or use of this software
     19  *    must display the following acknowledgement:
     20  *	This product includes software developed by the NetBSD
     21  *	Foundation, Inc. and its contributors.
     22  * 4. Neither the name of The NetBSD Foundation nor the names of its
     23  *    contributors may be used to endorse or promote products derived
     24  *    from this software without specific prior written permission.
     25  *
     26  * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS
     27  * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
     28  * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
     29  * PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS
     30  * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
     31  * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
     32  * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
     33  * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
     34  * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
     35  * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
     36  * POSSIBILITY OF SUCH DAMAGE.
     37  */
     38 
     39 #include <sys/cdefs.h>
     40 __KERNEL_RCSID(0, "$NetBSD: init_sysctl.c,v 1.129 2008/04/02 10:53:23 xtraeme Exp $");
     41 
     42 #include "opt_sysv.h"
     43 #include "opt_posix.h"
     44 #include "opt_compat_netbsd32.h"
     45 #include "pty.h"
     46 #include "rnd.h"
     47 
     48 #include <sys/types.h>
     49 #include <sys/param.h>
     50 #include <sys/sysctl.h>
     51 #include <sys/cpu.h>
     52 #include <sys/errno.h>
     53 #include <sys/systm.h>
     54 #include <sys/kernel.h>
     55 #include <sys/unistd.h>
     56 #include <sys/disklabel.h>
     57 #include <sys/rnd.h>
     58 #include <sys/vnode.h>
     59 #include <sys/mount.h>
     60 #include <sys/namei.h>
     61 #include <sys/msgbuf.h>
     62 #include <dev/cons.h>
     63 #include <sys/socketvar.h>
     64 #include <sys/file.h>
     65 #include <sys/filedesc.h>
     66 #include <sys/tty.h>
     67 #include <sys/malloc.h>
     68 #include <sys/resource.h>
     69 #include <sys/resourcevar.h>
     70 #include <sys/exec.h>
     71 #include <sys/conf.h>
     72 #include <sys/device.h>
     73 #include <sys/stat.h>
     74 #include <sys/kauth.h>
     75 #include <sys/ktrace.h>
     76 
     77 #ifdef COMPAT_NETBSD32
     78 #include <compat/netbsd32/netbsd32.h>
     79 #endif
     80 
     81 #include <sys/cpu.h>
     82 
     83 /* XXX this should not be here */
     84 int security_setidcore_dump;
     85 char security_setidcore_path[MAXPATHLEN] = "/var/crash/%n.core";
     86 uid_t security_setidcore_owner = 0;
     87 gid_t security_setidcore_group = 0;
     88 mode_t security_setidcore_mode = (S_IRUSR|S_IWUSR);
     89 
     90 static const u_int sysctl_flagmap[] = {
     91 	PK_ADVLOCK, P_ADVLOCK,
     92 	PK_EXEC, P_EXEC,
     93 	PK_NOCLDWAIT, P_NOCLDWAIT,
     94 	PK_32, P_32,
     95 	PK_CLDSIGIGN, P_CLDSIGIGN,
     96 	PK_SUGID, P_SUGID,
     97 	0
     98 };
     99 
    100 static const u_int sysctl_sflagmap[] = {
    101 	PS_NOCLDSTOP, P_NOCLDSTOP,
    102 	PS_PPWAIT, P_PPWAIT,
    103 	PS_WEXIT, P_WEXIT,
    104 	PS_STOPFORK, P_STOPFORK,
    105 	PS_STOPEXEC, P_STOPEXEC,
    106 	PS_STOPEXIT, P_STOPEXIT,
    107 	0
    108 };
    109 
    110 static const u_int sysctl_slflagmap[] = {
    111 	PSL_TRACED, P_TRACED,
    112 	PSL_FSTRACE, P_FSTRACE,
    113 	PSL_CHTRACED, P_CHTRACED,
    114 	PSL_SYSCALL, P_SYSCALL,
    115 	0
    116 };
    117 
    118 static const u_int sysctl_lflagmap[] = {
    119 	PL_CONTROLT, P_CONTROLT,
    120 	0
    121 };
    122 
    123 static const u_int sysctl_stflagmap[] = {
    124 	PST_PROFIL, P_PROFIL,
    125 	0
    126 
    127 };
    128 
    129 static const u_int sysctl_lwpflagmap[] = {
    130 	LW_INMEM, P_INMEM,
    131 	LW_SINTR, P_SINTR,
    132 	LW_SYSTEM, P_SYSTEM,
    133 	0
    134 };
    135 
    136 static const u_int sysctl_lwpprflagmap[] = {
    137 	LPR_DETACHED, L_DETACHED,
    138 	0
    139 };
    140 
    141 /*
    142  * try over estimating by 5 procs/lwps
    143  */
    144 #define KERN_PROCSLOP	(5 * sizeof(struct kinfo_proc))
    145 #define KERN_LWPSLOP	(5 * sizeof(struct kinfo_lwp))
    146 
    147 static int dcopyout(struct lwp *, const void *, void *, size_t);
    148 
    149 static int
    150 dcopyout(struct lwp *l, const void *kaddr, void *uaddr, size_t len)
    151 {
    152 	int error;
    153 
    154 	error = copyout(kaddr, uaddr, len);
    155 	ktrmibio(-1, UIO_READ, uaddr, len, error);
    156 
    157 	return error;
    158 }
    159 
    160 #ifdef DIAGNOSTIC
    161 static int sysctl_kern_trigger_panic(SYSCTLFN_PROTO);
    162 #endif
    163 static int sysctl_kern_maxvnodes(SYSCTLFN_PROTO);
    164 static int sysctl_kern_rtc_offset(SYSCTLFN_PROTO);
    165 static int sysctl_kern_maxproc(SYSCTLFN_PROTO);
    166 static int sysctl_kern_hostid(SYSCTLFN_PROTO);
    167 static int sysctl_setlen(SYSCTLFN_PROTO);
    168 static int sysctl_kern_clockrate(SYSCTLFN_PROTO);
    169 static int sysctl_kern_file(SYSCTLFN_PROTO);
    170 static int sysctl_msgbuf(SYSCTLFN_PROTO);
    171 static int sysctl_kern_defcorename(SYSCTLFN_PROTO);
    172 static int sysctl_kern_cptime(SYSCTLFN_PROTO);
    173 #if NPTY > 0
    174 static int sysctl_kern_maxptys(SYSCTLFN_PROTO);
    175 #endif /* NPTY > 0 */
    176 static int sysctl_kern_sbmax(SYSCTLFN_PROTO);
    177 static int sysctl_kern_urnd(SYSCTLFN_PROTO);
    178 static int sysctl_kern_arnd(SYSCTLFN_PROTO);
    179 static int sysctl_kern_lwp(SYSCTLFN_PROTO);
    180 static int sysctl_kern_forkfsleep(SYSCTLFN_PROTO);
    181 static int sysctl_kern_root_partition(SYSCTLFN_PROTO);
    182 static int sysctl_kern_drivers(SYSCTLFN_PROTO);
    183 static int sysctl_kern_file2(SYSCTLFN_PROTO);
    184 static int sysctl_security_setidcore(SYSCTLFN_PROTO);
    185 static int sysctl_security_setidcorename(SYSCTLFN_PROTO);
    186 static int sysctl_kern_cpid(SYSCTLFN_PROTO);
    187 static int sysctl_doeproc(SYSCTLFN_PROTO);
    188 static int sysctl_kern_proc_args(SYSCTLFN_PROTO);
    189 static int sysctl_hw_usermem(SYSCTLFN_PROTO);
    190 static int sysctl_hw_cnmagic(SYSCTLFN_PROTO);
    191 
    192 static u_int sysctl_map_flags(const u_int *, u_int);
    193 static void fill_kproc2(struct proc *, struct kinfo_proc2 *);
    194 static void fill_lwp(struct lwp *l, struct kinfo_lwp *kl);
    195 static void fill_file(struct kinfo_file *, const file_t *, const fdfile_t *,
    196 		      int, pid_t);
    197 
    198 /*
    199  * ********************************************************************
    200  * section 1: setup routines
    201  * ********************************************************************
    202  * These functions are stuffed into a link set for sysctl setup
    203  * functions. They're never called or referenced from anywhere else.
    204  * ********************************************************************
    205  */
    206 
    207 /*
    208  * sets up the base nodes...
    209  */
    210 SYSCTL_SETUP(sysctl_root_setup, "sysctl base setup")
    211 {
    212 
    213 	sysctl_createv(clog, 0, NULL, NULL,
    214 		       CTLFLAG_PERMANENT,
    215 		       CTLTYPE_NODE, "kern",
    216 		       SYSCTL_DESCR("High kernel"),
    217 		       NULL, 0, NULL, 0,
    218 		       CTL_KERN, CTL_EOL);
    219 	sysctl_createv(clog, 0, NULL, NULL,
    220 		       CTLFLAG_PERMANENT,
    221 		       CTLTYPE_NODE, "vm",
    222 		       SYSCTL_DESCR("Virtual memory"),
    223 		       NULL, 0, NULL, 0,
    224 		       CTL_VM, CTL_EOL);
    225 	sysctl_createv(clog, 0, NULL, NULL,
    226 		       CTLFLAG_PERMANENT,
    227 		       CTLTYPE_NODE, "vfs",
    228 		       SYSCTL_DESCR("Filesystem"),
    229 		       NULL, 0, NULL, 0,
    230 		       CTL_VFS, CTL_EOL);
    231 	sysctl_createv(clog, 0, NULL, NULL,
    232 		       CTLFLAG_PERMANENT,
    233 		       CTLTYPE_NODE, "net",
    234 		       SYSCTL_DESCR("Networking"),
    235 		       NULL, 0, NULL, 0,
    236 		       CTL_NET, CTL_EOL);
    237 	sysctl_createv(clog, 0, NULL, NULL,
    238 		       CTLFLAG_PERMANENT,
    239 		       CTLTYPE_NODE, "debug",
    240 		       SYSCTL_DESCR("Debugging"),
    241 		       NULL, 0, NULL, 0,
    242 		       CTL_DEBUG, CTL_EOL);
    243 	sysctl_createv(clog, 0, NULL, NULL,
    244 		       CTLFLAG_PERMANENT,
    245 		       CTLTYPE_NODE, "hw",
    246 		       SYSCTL_DESCR("Generic CPU, I/O"),
    247 		       NULL, 0, NULL, 0,
    248 		       CTL_HW, CTL_EOL);
    249 	sysctl_createv(clog, 0, NULL, NULL,
    250 		       CTLFLAG_PERMANENT,
    251 		       CTLTYPE_NODE, "machdep",
    252 		       SYSCTL_DESCR("Machine dependent"),
    253 		       NULL, 0, NULL, 0,
    254 		       CTL_MACHDEP, CTL_EOL);
    255 	/*
    256 	 * this node is inserted so that the sysctl nodes in libc can
    257 	 * operate.
    258 	 */
    259 	sysctl_createv(clog, 0, NULL, NULL,
    260 		       CTLFLAG_PERMANENT,
    261 		       CTLTYPE_NODE, "user",
    262 		       SYSCTL_DESCR("User-level"),
    263 		       NULL, 0, NULL, 0,
    264 		       CTL_USER, CTL_EOL);
    265 	sysctl_createv(clog, 0, NULL, NULL,
    266 		       CTLFLAG_PERMANENT,
    267 		       CTLTYPE_NODE, "ddb",
    268 		       SYSCTL_DESCR("In-kernel debugger"),
    269 		       NULL, 0, NULL, 0,
    270 		       CTL_DDB, CTL_EOL);
    271 	sysctl_createv(clog, 0, NULL, NULL,
    272 		       CTLFLAG_PERMANENT,
    273 		       CTLTYPE_NODE, "proc",
    274 		       SYSCTL_DESCR("Per-process"),
    275 		       NULL, 0, NULL, 0,
    276 		       CTL_PROC, CTL_EOL);
    277 	sysctl_createv(clog, 0, NULL, NULL,
    278 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
    279 		       CTLTYPE_NODE, "vendor",
    280 		       SYSCTL_DESCR("Vendor specific"),
    281 		       NULL, 0, NULL, 0,
    282 		       CTL_VENDOR, CTL_EOL);
    283 	sysctl_createv(clog, 0, NULL, NULL,
    284 		       CTLFLAG_PERMANENT,
    285 		       CTLTYPE_NODE, "emul",
    286 		       SYSCTL_DESCR("Emulation settings"),
    287 		       NULL, 0, NULL, 0,
    288 		       CTL_EMUL, CTL_EOL);
    289 	sysctl_createv(clog, 0, NULL, NULL,
    290 		       CTLFLAG_PERMANENT,
    291 		       CTLTYPE_NODE, "security",
    292 		       SYSCTL_DESCR("Security"),
    293 		       NULL, 0, NULL, 0,
    294 		       CTL_SECURITY, CTL_EOL);
    295 }
    296 
    297 /*
    298  * this setup routine is a replacement for kern_sysctl()
    299  */
    300 SYSCTL_SETUP(sysctl_kern_setup, "sysctl kern subtree setup")
    301 {
    302 	extern int kern_logsigexit;	/* defined in kern/kern_sig.c */
    303 	extern fixpt_t ccpu;		/* defined in kern/kern_synch.c */
    304 	extern int dumponpanic;		/* defined in kern/subr_prf.c */
    305 	const struct sysctlnode *rnode;
    306 
    307 	sysctl_createv(clog, 0, NULL, NULL,
    308 		       CTLFLAG_PERMANENT,
    309 		       CTLTYPE_NODE, "kern", NULL,
    310 		       NULL, 0, NULL, 0,
    311 		       CTL_KERN, CTL_EOL);
    312 
    313 	sysctl_createv(clog, 0, NULL, NULL,
    314 		       CTLFLAG_PERMANENT,
    315 		       CTLTYPE_STRING, "ostype",
    316 		       SYSCTL_DESCR("Operating system type"),
    317 		       NULL, 0, &ostype, 0,
    318 		       CTL_KERN, KERN_OSTYPE, CTL_EOL);
    319 	sysctl_createv(clog, 0, NULL, NULL,
    320 		       CTLFLAG_PERMANENT,
    321 		       CTLTYPE_STRING, "osrelease",
    322 		       SYSCTL_DESCR("Operating system release"),
    323 		       NULL, 0, &osrelease, 0,
    324 		       CTL_KERN, KERN_OSRELEASE, CTL_EOL);
    325 	sysctl_createv(clog, 0, NULL, NULL,
    326 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    327 		       CTLTYPE_INT, "osrevision",
    328 		       SYSCTL_DESCR("Operating system revision"),
    329 		       NULL, __NetBSD_Version__, NULL, 0,
    330 		       CTL_KERN, KERN_OSREV, CTL_EOL);
    331 	sysctl_createv(clog, 0, NULL, NULL,
    332 		       CTLFLAG_PERMANENT,
    333 		       CTLTYPE_STRING, "version",
    334 		       SYSCTL_DESCR("Kernel version"),
    335 		       NULL, 0, &version, 0,
    336 		       CTL_KERN, KERN_VERSION, CTL_EOL);
    337 	sysctl_createv(clog, 0, NULL, NULL,
    338 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
    339 		       CTLTYPE_INT, "maxvnodes",
    340 		       SYSCTL_DESCR("Maximum number of vnodes"),
    341 		       sysctl_kern_maxvnodes, 0, NULL, 0,
    342 		       CTL_KERN, KERN_MAXVNODES, CTL_EOL);
    343 	sysctl_createv(clog, 0, NULL, NULL,
    344 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
    345 		       CTLTYPE_INT, "maxproc",
    346 		       SYSCTL_DESCR("Maximum number of simultaneous processes"),
    347 		       sysctl_kern_maxproc, 0, NULL, 0,
    348 		       CTL_KERN, KERN_MAXPROC, CTL_EOL);
    349 	sysctl_createv(clog, 0, NULL, NULL,
    350 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
    351 		       CTLTYPE_INT, "maxfiles",
    352 		       SYSCTL_DESCR("Maximum number of open files"),
    353 		       NULL, 0, &maxfiles, 0,
    354 		       CTL_KERN, KERN_MAXFILES, CTL_EOL);
    355 	sysctl_createv(clog, 0, NULL, NULL,
    356 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    357 		       CTLTYPE_INT, "argmax",
    358 		       SYSCTL_DESCR("Maximum number of bytes of arguments to "
    359 				    "execve(2)"),
    360 		       NULL, ARG_MAX, NULL, 0,
    361 		       CTL_KERN, KERN_ARGMAX, CTL_EOL);
    362 	sysctl_createv(clog, 0, NULL, NULL,
    363 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
    364 		       CTLTYPE_STRING, "hostname",
    365 		       SYSCTL_DESCR("System hostname"),
    366 		       sysctl_setlen, 0, &hostname, MAXHOSTNAMELEN,
    367 		       CTL_KERN, KERN_HOSTNAME, CTL_EOL);
    368 	sysctl_createv(clog, 0, NULL, NULL,
    369 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE|CTLFLAG_HEX,
    370 		       CTLTYPE_INT, "hostid",
    371 		       SYSCTL_DESCR("System host ID number"),
    372 		       sysctl_kern_hostid, 0, NULL, 0,
    373 		       CTL_KERN, KERN_HOSTID, CTL_EOL);
    374 	sysctl_createv(clog, 0, NULL, NULL,
    375 		       CTLFLAG_PERMANENT,
    376 		       CTLTYPE_STRUCT, "clockrate",
    377 		       SYSCTL_DESCR("Kernel clock rates"),
    378 		       sysctl_kern_clockrate, 0, NULL,
    379 		       sizeof(struct clockinfo),
    380 		       CTL_KERN, KERN_CLOCKRATE, CTL_EOL);
    381 	sysctl_createv(clog, 0, NULL, NULL,
    382 		       CTLFLAG_PERMANENT,
    383 		       CTLTYPE_INT, "hardclock_ticks",
    384 		       SYSCTL_DESCR("Number of hardclock ticks"),
    385 		       NULL, 0, &hardclock_ticks, sizeof(hardclock_ticks),
    386 		       CTL_KERN, KERN_HARDCLOCK_TICKS, CTL_EOL);
    387 	sysctl_createv(clog, 0, NULL, NULL,
    388 		       CTLFLAG_PERMANENT,
    389 		       CTLTYPE_STRUCT, "vnode",
    390 		       SYSCTL_DESCR("System vnode table"),
    391 		       sysctl_kern_vnode, 0, NULL, 0,
    392 		       CTL_KERN, KERN_VNODE, CTL_EOL);
    393 	sysctl_createv(clog, 0, NULL, NULL,
    394 		       CTLFLAG_PERMANENT,
    395 		       CTLTYPE_STRUCT, "file",
    396 		       SYSCTL_DESCR("System open file table"),
    397 		       sysctl_kern_file, 0, NULL, 0,
    398 		       CTL_KERN, KERN_FILE, CTL_EOL);
    399 #ifndef GPROF
    400 	sysctl_createv(clog, 0, NULL, NULL,
    401 		       CTLFLAG_PERMANENT,
    402 		       CTLTYPE_NODE, "profiling",
    403 		       SYSCTL_DESCR("Profiling information (not available)"),
    404 		       sysctl_notavail, 0, NULL, 0,
    405 		       CTL_KERN, KERN_PROF, CTL_EOL);
    406 #endif
    407 	sysctl_createv(clog, 0, NULL, NULL,
    408 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    409 		       CTLTYPE_INT, "posix1version",
    410 		       SYSCTL_DESCR("Version of ISO/IEC 9945 (POSIX 1003.1) "
    411 				    "with which the operating system attempts "
    412 				    "to comply"),
    413 		       NULL, _POSIX_VERSION, NULL, 0,
    414 		       CTL_KERN, KERN_POSIX1, CTL_EOL);
    415 	sysctl_createv(clog, 0, NULL, NULL,
    416 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    417 		       CTLTYPE_INT, "ngroups",
    418 		       SYSCTL_DESCR("Maximum number of supplemental groups"),
    419 		       NULL, NGROUPS_MAX, NULL, 0,
    420 		       CTL_KERN, KERN_NGROUPS, CTL_EOL);
    421 	sysctl_createv(clog, 0, NULL, NULL,
    422 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    423 		       CTLTYPE_INT, "job_control",
    424 		       SYSCTL_DESCR("Whether job control is available"),
    425 		       NULL, 1, NULL, 0,
    426 		       CTL_KERN, KERN_JOB_CONTROL, CTL_EOL);
    427 	sysctl_createv(clog, 0, NULL, NULL,
    428 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    429 		       CTLTYPE_INT, "saved_ids",
    430 		       SYSCTL_DESCR("Whether POSIX saved set-group/user ID is "
    431 				    "available"), NULL,
    432 #ifdef _POSIX_SAVED_IDS
    433 		       1,
    434 #else /* _POSIX_SAVED_IDS */
    435 		       0,
    436 #endif /* _POSIX_SAVED_IDS */
    437 		       NULL, 0, CTL_KERN, KERN_SAVED_IDS, CTL_EOL);
    438 	sysctl_createv(clog, 0, NULL, NULL,
    439 		       CTLFLAG_PERMANENT,
    440 		       CTLTYPE_STRUCT, "boottime",
    441 		       SYSCTL_DESCR("System boot time"),
    442 		       NULL, 0, &boottime, sizeof(boottime),
    443 		       CTL_KERN, KERN_BOOTTIME, CTL_EOL);
    444 	sysctl_createv(clog, 0, NULL, NULL,
    445 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
    446 		       CTLTYPE_STRING, "domainname",
    447 		       SYSCTL_DESCR("YP domain name"),
    448 		       sysctl_setlen, 0, &domainname, MAXHOSTNAMELEN,
    449 		       CTL_KERN, KERN_DOMAINNAME, CTL_EOL);
    450 	sysctl_createv(clog, 0, NULL, NULL,
    451 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    452 		       CTLTYPE_INT, "maxpartitions",
    453 		       SYSCTL_DESCR("Maximum number of partitions allowed per "
    454 				    "disk"),
    455 		       NULL, MAXPARTITIONS, NULL, 0,
    456 		       CTL_KERN, KERN_MAXPARTITIONS, CTL_EOL);
    457 	sysctl_createv(clog, 0, NULL, NULL,
    458 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    459 		       CTLTYPE_INT, "rawpartition",
    460 		       SYSCTL_DESCR("Raw partition of a disk"),
    461 		       NULL, RAW_PART, NULL, 0,
    462 		       CTL_KERN, KERN_RAWPARTITION, CTL_EOL);
    463 	sysctl_createv(clog, 0, NULL, NULL,
    464 		       CTLFLAG_PERMANENT,
    465 		       CTLTYPE_STRUCT, "timex", NULL,
    466 		       sysctl_notavail, 0, NULL, 0,
    467 		       CTL_KERN, KERN_TIMEX, CTL_EOL);
    468 	sysctl_createv(clog, 0, NULL, NULL,
    469 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
    470 		       CTLTYPE_INT, "rtc_offset",
    471 		       SYSCTL_DESCR("Offset of real time clock from UTC in "
    472 				    "minutes"),
    473 		       sysctl_kern_rtc_offset, 0, &rtc_offset, 0,
    474 		       CTL_KERN, KERN_RTC_OFFSET, CTL_EOL);
    475 	sysctl_createv(clog, 0, NULL, NULL,
    476 		       CTLFLAG_PERMANENT,
    477 		       CTLTYPE_STRING, "root_device",
    478 		       SYSCTL_DESCR("Name of the root device"),
    479 		       sysctl_root_device, 0, NULL, 0,
    480 		       CTL_KERN, KERN_ROOT_DEVICE, CTL_EOL);
    481 	sysctl_createv(clog, 0, NULL, NULL,
    482 		       CTLFLAG_PERMANENT,
    483 		       CTLTYPE_INT, "msgbufsize",
    484 		       SYSCTL_DESCR("Size of the kernel message buffer"),
    485 		       sysctl_msgbuf, 0, NULL, 0,
    486 		       CTL_KERN, KERN_MSGBUFSIZE, CTL_EOL);
    487 	sysctl_createv(clog, 0, NULL, NULL,
    488 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    489 		       CTLTYPE_INT, "fsync",
    490 		       SYSCTL_DESCR("Whether the POSIX 1003.1b File "
    491 				    "Synchronization Option is available on "
    492 				    "this system"),
    493 		       NULL, 1, NULL, 0,
    494 		       CTL_KERN, KERN_FSYNC, CTL_EOL);
    495 	sysctl_createv(clog, 0, NULL, NULL,
    496 		       CTLFLAG_PERMANENT,
    497 		       CTLTYPE_NODE, "ipc",
    498 		       SYSCTL_DESCR("SysV IPC options"),
    499 		       NULL, 0, NULL, 0,
    500 		       CTL_KERN, KERN_SYSVIPC, CTL_EOL);
    501 	sysctl_createv(clog, 0, NULL, NULL,
    502 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    503 		       CTLTYPE_INT, "sysvmsg",
    504 		       SYSCTL_DESCR("System V style message support available"),
    505 		       NULL,
    506 #ifdef SYSVMSG
    507 		       1,
    508 #else /* SYSVMSG */
    509 		       0,
    510 #endif /* SYSVMSG */
    511 		       NULL, 0, CTL_KERN, KERN_SYSVIPC, KERN_SYSVIPC_MSG, CTL_EOL);
    512 	sysctl_createv(clog, 0, NULL, NULL,
    513 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    514 		       CTLTYPE_INT, "sysvsem",
    515 		       SYSCTL_DESCR("System V style semaphore support "
    516 				    "available"), NULL,
    517 #ifdef SYSVSEM
    518 		       1,
    519 #else /* SYSVSEM */
    520 		       0,
    521 #endif /* SYSVSEM */
    522 		       NULL, 0, CTL_KERN, KERN_SYSVIPC, KERN_SYSVIPC_SEM, CTL_EOL);
    523 	sysctl_createv(clog, 0, NULL, NULL,
    524 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    525 		       CTLTYPE_INT, "sysvshm",
    526 		       SYSCTL_DESCR("System V style shared memory support "
    527 				    "available"), NULL,
    528 #ifdef SYSVSHM
    529 		       1,
    530 #else /* SYSVSHM */
    531 		       0,
    532 #endif /* SYSVSHM */
    533 		       NULL, 0, CTL_KERN, KERN_SYSVIPC, KERN_SYSVIPC_SHM, CTL_EOL);
    534 	sysctl_createv(clog, 0, NULL, NULL,
    535 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    536 		       CTLTYPE_INT, "synchronized_io",
    537 		       SYSCTL_DESCR("Whether the POSIX 1003.1b Synchronized "
    538 				    "I/O Option is available on this system"),
    539 		       NULL, 1, NULL, 0,
    540 		       CTL_KERN, KERN_SYNCHRONIZED_IO, CTL_EOL);
    541 	sysctl_createv(clog, 0, NULL, NULL,
    542 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    543 		       CTLTYPE_INT, "iov_max",
    544 		       SYSCTL_DESCR("Maximum number of iovec structures per "
    545 				    "process"),
    546 		       NULL, IOV_MAX, NULL, 0,
    547 		       CTL_KERN, KERN_IOV_MAX, CTL_EOL);
    548 	sysctl_createv(clog, 0, NULL, NULL,
    549 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    550 		       CTLTYPE_INT, "mapped_files",
    551 		       SYSCTL_DESCR("Whether the POSIX 1003.1b Memory Mapped "
    552 				    "Files Option is available on this system"),
    553 		       NULL, 1, NULL, 0,
    554 		       CTL_KERN, KERN_MAPPED_FILES, CTL_EOL);
    555 	sysctl_createv(clog, 0, NULL, NULL,
    556 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    557 		       CTLTYPE_INT, "memlock",
    558 		       SYSCTL_DESCR("Whether the POSIX 1003.1b Process Memory "
    559 				    "Locking Option is available on this "
    560 				    "system"),
    561 		       NULL, 1, NULL, 0,
    562 		       CTL_KERN, KERN_MEMLOCK, CTL_EOL);
    563 	sysctl_createv(clog, 0, NULL, NULL,
    564 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    565 		       CTLTYPE_INT, "memlock_range",
    566 		       SYSCTL_DESCR("Whether the POSIX 1003.1b Range Memory "
    567 				    "Locking Option is available on this "
    568 				    "system"),
    569 		       NULL, 1, NULL, 0,
    570 		       CTL_KERN, KERN_MEMLOCK_RANGE, CTL_EOL);
    571 	sysctl_createv(clog, 0, NULL, NULL,
    572 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    573 		       CTLTYPE_INT, "memory_protection",
    574 		       SYSCTL_DESCR("Whether the POSIX 1003.1b Memory "
    575 				    "Protection Option is available on this "
    576 				    "system"),
    577 		       NULL, 1, NULL, 0,
    578 		       CTL_KERN, KERN_MEMORY_PROTECTION, CTL_EOL);
    579 	sysctl_createv(clog, 0, NULL, NULL,
    580 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    581 		       CTLTYPE_INT, "login_name_max",
    582 		       SYSCTL_DESCR("Maximum login name length"),
    583 		       NULL, LOGIN_NAME_MAX, NULL, 0,
    584 		       CTL_KERN, KERN_LOGIN_NAME_MAX, CTL_EOL);
    585 	sysctl_createv(clog, 0, NULL, NULL,
    586 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
    587 		       CTLTYPE_STRING, "defcorename",
    588 		       SYSCTL_DESCR("Default core file name"),
    589 		       sysctl_kern_defcorename, 0, defcorename, MAXPATHLEN,
    590 		       CTL_KERN, KERN_DEFCORENAME, CTL_EOL);
    591 	sysctl_createv(clog, 0, NULL, NULL,
    592 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
    593 		       CTLTYPE_INT, "logsigexit",
    594 		       SYSCTL_DESCR("Log process exit when caused by signals"),
    595 		       NULL, 0, &kern_logsigexit, 0,
    596 		       CTL_KERN, KERN_LOGSIGEXIT, CTL_EOL);
    597 	sysctl_createv(clog, 0, NULL, NULL,
    598 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    599 		       CTLTYPE_INT, "fscale",
    600 		       SYSCTL_DESCR("Kernel fixed-point scale factor"),
    601 		       NULL, FSCALE, NULL, 0,
    602 		       CTL_KERN, KERN_FSCALE, CTL_EOL);
    603 	sysctl_createv(clog, 0, NULL, NULL,
    604 		       CTLFLAG_PERMANENT,
    605 		       CTLTYPE_INT, "ccpu",
    606 		       SYSCTL_DESCR("Scheduler exponential decay value"),
    607 		       NULL, 0, &ccpu, 0,
    608 		       CTL_KERN, KERN_CCPU, CTL_EOL);
    609 	sysctl_createv(clog, 0, NULL, NULL,
    610 		       CTLFLAG_PERMANENT,
    611 		       CTLTYPE_STRUCT, "cp_time",
    612 		       SYSCTL_DESCR("Clock ticks spent in different CPU states"),
    613 		       sysctl_kern_cptime, 0, NULL, 0,
    614 		       CTL_KERN, KERN_CP_TIME, CTL_EOL);
    615 	sysctl_createv(clog, 0, NULL, NULL,
    616 		       CTLFLAG_PERMANENT,
    617 		       CTLTYPE_INT, "msgbuf",
    618 		       SYSCTL_DESCR("Kernel message buffer"),
    619 		       sysctl_msgbuf, 0, NULL, 0,
    620 		       CTL_KERN, KERN_MSGBUF, CTL_EOL);
    621 	sysctl_createv(clog, 0, NULL, NULL,
    622 		       CTLFLAG_PERMANENT,
    623 		       CTLTYPE_STRUCT, "consdev",
    624 		       SYSCTL_DESCR("Console device"),
    625 		       sysctl_consdev, 0, NULL, sizeof(dev_t),
    626 		       CTL_KERN, KERN_CONSDEV, CTL_EOL);
    627 #if NPTY > 0
    628 	sysctl_createv(clog, 0, NULL, NULL,
    629 		       CTLFLAG_PERMANENT,
    630 		       CTLTYPE_INT, "maxptys",
    631 		       SYSCTL_DESCR("Maximum number of pseudo-ttys"),
    632 		       sysctl_kern_maxptys, 0, NULL, 0,
    633 		       CTL_KERN, KERN_MAXPTYS, CTL_EOL);
    634 #endif /* NPTY > 0 */
    635 	sysctl_createv(clog, 0, NULL, NULL,
    636 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    637 		       CTLTYPE_INT, "maxphys",
    638 		       SYSCTL_DESCR("Maximum raw I/O transfer size"),
    639 		       NULL, MAXPHYS, NULL, 0,
    640 		       CTL_KERN, KERN_MAXPHYS, CTL_EOL);
    641 	sysctl_createv(clog, 0, NULL, NULL,
    642 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
    643 		       CTLTYPE_INT, "sbmax",
    644 		       SYSCTL_DESCR("Maximum socket buffer size"),
    645 		       sysctl_kern_sbmax, 0, NULL, 0,
    646 		       CTL_KERN, KERN_SBMAX, CTL_EOL);
    647 	sysctl_createv(clog, 0, NULL, NULL,
    648 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    649 		       CTLTYPE_INT, "monotonic_clock",
    650 		       SYSCTL_DESCR("Implementation version of the POSIX "
    651 				    "1003.1b Monotonic Clock Option"),
    652 		       /* XXX _POSIX_VERSION */
    653 		       NULL, _POSIX_MONOTONIC_CLOCK, NULL, 0,
    654 		       CTL_KERN, KERN_MONOTONIC_CLOCK, CTL_EOL);
    655 	sysctl_createv(clog, 0, NULL, NULL,
    656 		       CTLFLAG_PERMANENT,
    657 		       CTLTYPE_INT, "urandom",
    658 		       SYSCTL_DESCR("Random integer value"),
    659 		       sysctl_kern_urnd, 0, NULL, 0,
    660 		       CTL_KERN, KERN_URND, CTL_EOL);
    661 	sysctl_createv(clog, 0, NULL, NULL,
    662 		       CTLFLAG_PERMANENT,
    663 		       CTLTYPE_INT, "arandom",
    664 		       SYSCTL_DESCR("n bytes of random data"),
    665 		       sysctl_kern_arnd, 0, NULL, 0,
    666 		       CTL_KERN, KERN_ARND, CTL_EOL);
    667 	sysctl_createv(clog, 0, NULL, NULL,
    668 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    669 		       CTLTYPE_INT, "labelsector",
    670 		       SYSCTL_DESCR("Sector number containing the disklabel"),
    671 		       NULL, LABELSECTOR, NULL, 0,
    672 		       CTL_KERN, KERN_LABELSECTOR, CTL_EOL);
    673 	sysctl_createv(clog, 0, NULL, NULL,
    674 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    675 		       CTLTYPE_INT, "labeloffset",
    676 		       SYSCTL_DESCR("Offset of the disklabel within the "
    677 				    "sector"),
    678 		       NULL, LABELOFFSET, NULL, 0,
    679 		       CTL_KERN, KERN_LABELOFFSET, CTL_EOL);
    680 	sysctl_createv(clog, 0, NULL, NULL,
    681 		       CTLFLAG_PERMANENT,
    682 		       CTLTYPE_NODE, "lwp",
    683 		       SYSCTL_DESCR("System-wide LWP information"),
    684 		       sysctl_kern_lwp, 0, NULL, 0,
    685 		       CTL_KERN, KERN_LWP, CTL_EOL);
    686 	sysctl_createv(clog, 0, NULL, NULL,
    687 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
    688 		       CTLTYPE_INT, "forkfsleep",
    689 		       SYSCTL_DESCR("Milliseconds to sleep on fork failure due "
    690 				    "to process limits"),
    691 		       sysctl_kern_forkfsleep, 0, NULL, 0,
    692 		       CTL_KERN, KERN_FORKFSLEEP, CTL_EOL);
    693 	sysctl_createv(clog, 0, NULL, NULL,
    694 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    695 		       CTLTYPE_INT, "posix_threads",
    696 		       SYSCTL_DESCR("Version of IEEE Std 1003.1 and its "
    697 				    "Threads option to which the system "
    698 				    "attempts to conform"),
    699 		       /* XXX _POSIX_VERSION */
    700 		       NULL, _POSIX_THREADS, NULL, 0,
    701 		       CTL_KERN, KERN_POSIX_THREADS, CTL_EOL);
    702 	sysctl_createv(clog, 0, NULL, NULL,
    703 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    704 		       CTLTYPE_INT, "posix_semaphores",
    705 		       SYSCTL_DESCR("Version of IEEE Std 1003.1 and its "
    706 				    "Semaphores option to which the system "
    707 				    "attempts to conform"), NULL,
    708 #ifdef P1003_1B_SEMAPHORE
    709 		       200112,
    710 #else /* P1003_1B_SEMAPHORE */
    711 		       0,
    712 #endif /* P1003_1B_SEMAPHORE */
    713 		       NULL, 0, CTL_KERN, KERN_POSIX_SEMAPHORES, CTL_EOL);
    714 	sysctl_createv(clog, 0, NULL, NULL,
    715 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    716 		       CTLTYPE_INT, "posix_barriers",
    717 		       SYSCTL_DESCR("Version of IEEE Std 1003.1 and its "
    718 				    "Barriers option to which the system "
    719 				    "attempts to conform"),
    720 		       /* XXX _POSIX_VERSION */
    721 		       NULL, _POSIX_BARRIERS, NULL, 0,
    722 		       CTL_KERN, KERN_POSIX_BARRIERS, CTL_EOL);
    723 	sysctl_createv(clog, 0, NULL, NULL,
    724 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    725 		       CTLTYPE_INT, "posix_timers",
    726 		       SYSCTL_DESCR("Version of IEEE Std 1003.1 and its "
    727 				    "Timers option to which the system "
    728 				    "attempts to conform"),
    729 		       /* XXX _POSIX_VERSION */
    730 		       NULL, _POSIX_TIMERS, NULL, 0,
    731 		       CTL_KERN, KERN_POSIX_TIMERS, CTL_EOL);
    732 	sysctl_createv(clog, 0, NULL, NULL,
    733 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    734 		       CTLTYPE_INT, "posix_spin_locks",
    735 		       SYSCTL_DESCR("Version of IEEE Std 1003.1 and its Spin "
    736 				    "Locks option to which the system attempts "
    737 				    "to conform"),
    738 		       /* XXX _POSIX_VERSION */
    739 		       NULL, _POSIX_SPIN_LOCKS, NULL, 0,
    740 		       CTL_KERN, KERN_POSIX_SPIN_LOCKS, CTL_EOL);
    741 	sysctl_createv(clog, 0, NULL, NULL,
    742 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    743 		       CTLTYPE_INT, "posix_reader_writer_locks",
    744 		       SYSCTL_DESCR("Version of IEEE Std 1003.1 and its "
    745 				    "Read-Write Locks option to which the "
    746 				    "system attempts to conform"),
    747 		       /* XXX _POSIX_VERSION */
    748 		       NULL, _POSIX_READER_WRITER_LOCKS, NULL, 0,
    749 		       CTL_KERN, KERN_POSIX_READER_WRITER_LOCKS, CTL_EOL);
    750 	sysctl_createv(clog, 0, NULL, NULL,
    751 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
    752 		       CTLTYPE_INT, "dump_on_panic",
    753 		       SYSCTL_DESCR("Perform a crash dump on system panic"),
    754 		       NULL, 0, &dumponpanic, 0,
    755 		       CTL_KERN, KERN_DUMP_ON_PANIC, CTL_EOL);
    756 #ifdef DIAGNOSTIC
    757 	sysctl_createv(clog, 0, NULL, NULL,
    758 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
    759 		       CTLTYPE_INT, "panic_now",
    760 		       SYSCTL_DESCR("Trigger a panic"),
    761 		       sysctl_kern_trigger_panic, 0, NULL, 0,
    762 		       CTL_KERN, CTL_CREATE, CTL_EOL);
    763 #endif
    764 	sysctl_createv(clog, 0, NULL, NULL,
    765 		       CTLFLAG_PERMANENT,
    766 		       CTLTYPE_INT, "root_partition",
    767 		       SYSCTL_DESCR("Root partition on the root device"),
    768 		       sysctl_kern_root_partition, 0, NULL, 0,
    769 		       CTL_KERN, KERN_ROOT_PARTITION, CTL_EOL);
    770 	sysctl_createv(clog, 0, NULL, NULL,
    771 		       CTLFLAG_PERMANENT,
    772 		       CTLTYPE_STRUCT, "drivers",
    773 		       SYSCTL_DESCR("List of all drivers with block and "
    774 				    "character device numbers"),
    775 		       sysctl_kern_drivers, 0, NULL, 0,
    776 		       CTL_KERN, KERN_DRIVERS, CTL_EOL);
    777 	sysctl_createv(clog, 0, NULL, NULL,
    778 		       CTLFLAG_PERMANENT,
    779 		       CTLTYPE_STRUCT, "file2",
    780 		       SYSCTL_DESCR("System open file table"),
    781 		       sysctl_kern_file2, 0, NULL, 0,
    782 		       CTL_KERN, KERN_FILE2, CTL_EOL);
    783 	sysctl_createv(clog, 0, NULL, NULL,
    784 		       CTLFLAG_PERMANENT,
    785 		       CTLTYPE_STRUCT, "cp_id",
    786 		       SYSCTL_DESCR("Mapping of CPU number to CPU id"),
    787 		       sysctl_kern_cpid, 0, NULL, 0,
    788 		       CTL_KERN, KERN_CP_ID, CTL_EOL);
    789 	sysctl_createv(clog, 0, NULL, &rnode,
    790 		       CTLFLAG_PERMANENT,
    791 		       CTLTYPE_NODE, "coredump",
    792 		       SYSCTL_DESCR("Coredump settings."),
    793 		       NULL, 0, NULL, 0,
    794 		       CTL_KERN, CTL_CREATE, CTL_EOL);
    795 	sysctl_createv(clog, 0, &rnode, &rnode,
    796 		       CTLFLAG_PERMANENT,
    797 		       CTLTYPE_NODE, "setid",
    798 		       SYSCTL_DESCR("Set-id processes' coredump settings."),
    799 		       NULL, 0, NULL, 0,
    800 		       CTL_CREATE, CTL_EOL);
    801 	sysctl_createv(clog, 0, &rnode, NULL,
    802 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
    803 		       CTLTYPE_INT, "dump",
    804 		       SYSCTL_DESCR("Allow set-id processes to dump core."),
    805 		       sysctl_security_setidcore, 0, &security_setidcore_dump,
    806 		       sizeof(security_setidcore_dump),
    807 		       CTL_CREATE, CTL_EOL);
    808 	sysctl_createv(clog, 0, &rnode, NULL,
    809 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
    810 		       CTLTYPE_STRING, "path",
    811 		       SYSCTL_DESCR("Path pattern for set-id coredumps."),
    812 		       sysctl_security_setidcorename, 0,
    813 		       &security_setidcore_path,
    814 		       sizeof(security_setidcore_path),
    815 		       CTL_CREATE, CTL_EOL);
    816 	sysctl_createv(clog, 0, &rnode, NULL,
    817 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
    818 		       CTLTYPE_INT, "owner",
    819 		       SYSCTL_DESCR("Owner id for set-id processes' cores."),
    820 		       sysctl_security_setidcore, 0, &security_setidcore_owner,
    821 		       0,
    822 		       CTL_CREATE, CTL_EOL);
    823 	sysctl_createv(clog, 0, &rnode, NULL,
    824 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
    825 		       CTLTYPE_INT, "group",
    826 		       SYSCTL_DESCR("Group id for set-id processes' cores."),
    827 		       sysctl_security_setidcore, 0, &security_setidcore_group,
    828 		       0,
    829 		       CTL_CREATE, CTL_EOL);
    830 	sysctl_createv(clog, 0, &rnode, NULL,
    831 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
    832 		       CTLTYPE_INT, "mode",
    833 		       SYSCTL_DESCR("Mode for set-id processes' cores."),
    834 		       sysctl_security_setidcore, 0, &security_setidcore_mode,
    835 		       0,
    836 		       CTL_CREATE, CTL_EOL);
    837 	sysctl_createv(clog, 0, NULL, NULL,
    838 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    839 		       CTLTYPE_INT, "no_sa_support",
    840 		       SYSCTL_DESCR("0 if the kernel supports SA, otherwise it doesn't"),
    841 		       NULL, 1, NULL, 0,
    842 		       CTL_KERN, CTL_CREATE, CTL_EOL);
    843 }
    844 
    845 SYSCTL_SETUP(sysctl_kern_proc_setup,
    846 	     "sysctl kern.proc/proc2/proc_args subtree setup")
    847 {
    848 
    849 	sysctl_createv(clog, 0, NULL, NULL,
    850 		       CTLFLAG_PERMANENT,
    851 		       CTLTYPE_NODE, "kern", NULL,
    852 		       NULL, 0, NULL, 0,
    853 		       CTL_KERN, CTL_EOL);
    854 
    855 	sysctl_createv(clog, 0, NULL, NULL,
    856 		       CTLFLAG_PERMANENT,
    857 		       CTLTYPE_NODE, "proc",
    858 		       SYSCTL_DESCR("System-wide process information"),
    859 		       sysctl_doeproc, 0, NULL, 0,
    860 		       CTL_KERN, KERN_PROC, CTL_EOL);
    861 	sysctl_createv(clog, 0, NULL, NULL,
    862 		       CTLFLAG_PERMANENT,
    863 		       CTLTYPE_NODE, "proc2",
    864 		       SYSCTL_DESCR("Machine-independent process information"),
    865 		       sysctl_doeproc, 0, NULL, 0,
    866 		       CTL_KERN, KERN_PROC2, CTL_EOL);
    867 	sysctl_createv(clog, 0, NULL, NULL,
    868 		       CTLFLAG_PERMANENT,
    869 		       CTLTYPE_NODE, "proc_args",
    870 		       SYSCTL_DESCR("Process argument information"),
    871 		       sysctl_kern_proc_args, 0, NULL, 0,
    872 		       CTL_KERN, KERN_PROC_ARGS, CTL_EOL);
    873 
    874 	/*
    875 	  "nodes" under these:
    876 
    877 	  KERN_PROC_ALL
    878 	  KERN_PROC_PID pid
    879 	  KERN_PROC_PGRP pgrp
    880 	  KERN_PROC_SESSION sess
    881 	  KERN_PROC_TTY tty
    882 	  KERN_PROC_UID uid
    883 	  KERN_PROC_RUID uid
    884 	  KERN_PROC_GID gid
    885 	  KERN_PROC_RGID gid
    886 
    887 	  all in all, probably not worth the effort...
    888 	*/
    889 }
    890 
    891 SYSCTL_SETUP(sysctl_hw_setup, "sysctl hw subtree setup")
    892 {
    893 	u_int u;
    894 	u_quad_t q;
    895 
    896 	sysctl_createv(clog, 0, NULL, NULL,
    897 		       CTLFLAG_PERMANENT,
    898 		       CTLTYPE_NODE, "hw", NULL,
    899 		       NULL, 0, NULL, 0,
    900 		       CTL_HW, CTL_EOL);
    901 
    902 	sysctl_createv(clog, 0, NULL, NULL,
    903 		       CTLFLAG_PERMANENT,
    904 		       CTLTYPE_STRING, "machine",
    905 		       SYSCTL_DESCR("Machine class"),
    906 		       NULL, 0, machine, 0,
    907 		       CTL_HW, HW_MACHINE, CTL_EOL);
    908 	sysctl_createv(clog, 0, NULL, NULL,
    909 		       CTLFLAG_PERMANENT,
    910 		       CTLTYPE_STRING, "model",
    911 		       SYSCTL_DESCR("Machine model"),
    912 		       NULL, 0, cpu_model, 0,
    913 		       CTL_HW, HW_MODEL, CTL_EOL);
    914 	sysctl_createv(clog, 0, NULL, NULL,
    915 		       CTLFLAG_PERMANENT,
    916 		       CTLTYPE_INT, "ncpu",
    917 		       SYSCTL_DESCR("Number of CPUs configured"),
    918 		       NULL, 0, &ncpu, 0,
    919 		       CTL_HW, HW_NCPU, CTL_EOL);
    920 	sysctl_createv(clog, 0, NULL, NULL,
    921 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    922 		       CTLTYPE_INT, "byteorder",
    923 		       SYSCTL_DESCR("System byte order"),
    924 		       NULL, BYTE_ORDER, NULL, 0,
    925 		       CTL_HW, HW_BYTEORDER, CTL_EOL);
    926 	u = ((u_int)physmem > (UINT_MAX / PAGE_SIZE)) ?
    927 		UINT_MAX : physmem * PAGE_SIZE;
    928 	sysctl_createv(clog, 0, NULL, NULL,
    929 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    930 		       CTLTYPE_INT, "physmem",
    931 		       SYSCTL_DESCR("Bytes of physical memory"),
    932 		       NULL, u, NULL, 0,
    933 		       CTL_HW, HW_PHYSMEM, CTL_EOL);
    934 	sysctl_createv(clog, 0, NULL, NULL,
    935 		       CTLFLAG_PERMANENT,
    936 		       CTLTYPE_INT, "usermem",
    937 		       SYSCTL_DESCR("Bytes of non-kernel memory"),
    938 		       sysctl_hw_usermem, 0, NULL, 0,
    939 		       CTL_HW, HW_USERMEM, CTL_EOL);
    940 	sysctl_createv(clog, 0, NULL, NULL,
    941 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    942 		       CTLTYPE_INT, "pagesize",
    943 		       SYSCTL_DESCR("Software page size"),
    944 		       NULL, PAGE_SIZE, NULL, 0,
    945 		       CTL_HW, HW_PAGESIZE, CTL_EOL);
    946 	sysctl_createv(clog, 0, NULL, NULL,
    947 		       CTLFLAG_PERMANENT,
    948 		       CTLTYPE_STRING, "machine_arch",
    949 		       SYSCTL_DESCR("Machine CPU class"),
    950 		       NULL, 0, machine_arch, 0,
    951 		       CTL_HW, HW_MACHINE_ARCH, CTL_EOL);
    952 	sysctl_createv(clog, 0, NULL, NULL,
    953 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    954 		       CTLTYPE_INT, "alignbytes",
    955 		       SYSCTL_DESCR("Alignment constraint for all possible "
    956 				    "data types"),
    957 		       NULL, ALIGNBYTES, NULL, 0,
    958 		       CTL_HW, HW_ALIGNBYTES, CTL_EOL);
    959 	sysctl_createv(clog, 0, NULL, NULL,
    960 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE|CTLFLAG_HEX,
    961 		       CTLTYPE_STRING, "cnmagic",
    962 		       SYSCTL_DESCR("Console magic key sequence"),
    963 		       sysctl_hw_cnmagic, 0, NULL, CNS_LEN,
    964 		       CTL_HW, HW_CNMAGIC, CTL_EOL);
    965 	q = (u_quad_t)physmem * PAGE_SIZE;
    966 	sysctl_createv(clog, 0, NULL, NULL,
    967 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    968 		       CTLTYPE_QUAD, "physmem64",
    969 		       SYSCTL_DESCR("Bytes of physical memory"),
    970 		       NULL, q, NULL, 0,
    971 		       CTL_HW, HW_PHYSMEM64, CTL_EOL);
    972 	sysctl_createv(clog, 0, NULL, NULL,
    973 		       CTLFLAG_PERMANENT,
    974 		       CTLTYPE_QUAD, "usermem64",
    975 		       SYSCTL_DESCR("Bytes of non-kernel memory"),
    976 		       sysctl_hw_usermem, 0, NULL, 0,
    977 		       CTL_HW, HW_USERMEM64, CTL_EOL);
    978 	sysctl_createv(clog, 0, NULL, NULL,
    979 		       CTLFLAG_PERMANENT,
    980 		       CTLTYPE_INT, "ncpuonline",
    981 		       SYSCTL_DESCR("Number of CPUs online"),
    982 		       NULL, 0, &ncpuonline, 0,
    983 		       CTL_HW, HW_NCPUONLINE, CTL_EOL);
    984 }
    985 
    986 #ifdef DEBUG
    987 /*
    988  * Debugging related system variables.
    989  */
    990 struct ctldebug /* debug0, */ /* debug1, */ debug2, debug3, debug4;
    991 struct ctldebug debug5, debug6, debug7, debug8, debug9;
    992 struct ctldebug debug10, debug11, debug12, debug13, debug14;
    993 struct ctldebug debug15, debug16, debug17, debug18, debug19;
    994 static struct ctldebug *debugvars[CTL_DEBUG_MAXID] = {
    995 	&debug0, &debug1, &debug2, &debug3, &debug4,
    996 	&debug5, &debug6, &debug7, &debug8, &debug9,
    997 	&debug10, &debug11, &debug12, &debug13, &debug14,
    998 	&debug15, &debug16, &debug17, &debug18, &debug19,
    999 };
   1000 
   1001 /*
   1002  * this setup routine is a replacement for debug_sysctl()
   1003  *
   1004  * note that it creates several nodes per defined debug variable
   1005  */
   1006 SYSCTL_SETUP(sysctl_debug_setup, "sysctl debug subtree setup")
   1007 {
   1008 	struct ctldebug *cdp;
   1009 	char nodename[20];
   1010 	int i;
   1011 
   1012 	/*
   1013 	 * two ways here:
   1014 	 *
   1015 	 * the "old" way (debug.name -> value) which was emulated by
   1016 	 * the sysctl(8) binary
   1017 	 *
   1018 	 * the new way, which the sysctl(8) binary was actually using
   1019 
   1020 	 node	debug
   1021 	 node	debug.0
   1022 	 string debug.0.name
   1023 	 int	debug.0.value
   1024 	 int	debug.name
   1025 
   1026 	 */
   1027 
   1028 	sysctl_createv(clog, 0, NULL, NULL,
   1029 		       CTLFLAG_PERMANENT,
   1030 		       CTLTYPE_NODE, "debug", NULL,
   1031 		       NULL, 0, NULL, 0,
   1032 		       CTL_DEBUG, CTL_EOL);
   1033 
   1034 	for (i = 0; i < CTL_DEBUG_MAXID; i++) {
   1035 		cdp = debugvars[i];
   1036 		if (cdp->debugname == NULL || cdp->debugvar == NULL)
   1037 			continue;
   1038 
   1039 		snprintf(nodename, sizeof(nodename), "debug%d", i);
   1040 		sysctl_createv(clog, 0, NULL, NULL,
   1041 			       CTLFLAG_PERMANENT|CTLFLAG_HIDDEN,
   1042 			       CTLTYPE_NODE, nodename, NULL,
   1043 			       NULL, 0, NULL, 0,
   1044 			       CTL_DEBUG, i, CTL_EOL);
   1045 		sysctl_createv(clog, 0, NULL, NULL,
   1046 			       CTLFLAG_PERMANENT|CTLFLAG_HIDDEN,
   1047 			       CTLTYPE_STRING, "name", NULL,
   1048 			       /*XXXUNCONST*/
   1049 			       NULL, 0, __UNCONST(cdp->debugname), 0,
   1050 			       CTL_DEBUG, i, CTL_DEBUG_NAME, CTL_EOL);
   1051 		sysctl_createv(clog, 0, NULL, NULL,
   1052 			       CTLFLAG_PERMANENT|CTLFLAG_HIDDEN,
   1053 			       CTLTYPE_INT, "value", NULL,
   1054 			       NULL, 0, cdp->debugvar, 0,
   1055 			       CTL_DEBUG, i, CTL_DEBUG_VALUE, CTL_EOL);
   1056 		sysctl_createv(clog, 0, NULL, NULL,
   1057 			       CTLFLAG_PERMANENT,
   1058 			       CTLTYPE_INT, cdp->debugname, NULL,
   1059 			       NULL, 0, cdp->debugvar, 0,
   1060 			       CTL_DEBUG, CTL_CREATE, CTL_EOL);
   1061 	}
   1062 }
   1063 #endif /* DEBUG */
   1064 
   1065 /*
   1066  * ********************************************************************
   1067  * section 2: private node-specific helper routines.
   1068  * ********************************************************************
   1069  */
   1070 
   1071 #ifdef DIAGNOSTIC
   1072 static int
   1073 sysctl_kern_trigger_panic(SYSCTLFN_ARGS)
   1074 {
   1075 	int newtrig, error;
   1076 	struct sysctlnode node;
   1077 
   1078 	newtrig = 0;
   1079 	node = *rnode;
   1080 	node.sysctl_data = &newtrig;
   1081 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
   1082 	if (error || newp == NULL)
   1083 		return (error);
   1084 
   1085 	if (newtrig != 0)
   1086 		panic("Panic triggered");
   1087 
   1088 	return (error);
   1089 }
   1090 #endif
   1091 
   1092 /*
   1093  * sysctl helper routine for kern.maxvnodes.  Drain vnodes if
   1094  * new value is lower than desiredvnodes and then calls reinit
   1095  * routines that needs to adjust to the new value.
   1096  */
   1097 static int
   1098 sysctl_kern_maxvnodes(SYSCTLFN_ARGS)
   1099 {
   1100 	int error, new_vnodes, old_vnodes;
   1101 	struct sysctlnode node;
   1102 
   1103 	new_vnodes = desiredvnodes;
   1104 	node = *rnode;
   1105 	node.sysctl_data = &new_vnodes;
   1106 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
   1107 	if (error || newp == NULL)
   1108 		return (error);
   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(&proclist_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(&proclist_lock);
   1691 
   1692 			mutex_enter(&p->p_smutex);
   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_smutex);
   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_smutex);
   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_smutex);
   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_smutex);
   1730 
   1731 			/* Drop reference to process. */
   1732 			mutex_enter(&proclist_lock);
   1733 			rw_exit(&p->p_reflock);
   1734 		}
   1735 		mutex_exit(&proclist_lock);
   1736 	} else {
   1737 		mutex_enter(&proclist_lock);
   1738 		p = p_find(pid, PFIND_LOCKED);
   1739 		if (p == NULL) {
   1740 			error = ESRCH;
   1741 			mutex_exit(&proclist_lock);
   1742 			goto cleanup;
   1743 		}
   1744 		/* Grab a hold on the process. */
   1745 		gotit = rw_tryenter(&p->p_reflock, RW_READER);
   1746 		mutex_exit(&proclist_lock);
   1747 		if (!gotit) {
   1748 			error = ESRCH;
   1749 			goto cleanup;
   1750 		}
   1751 
   1752 		mutex_enter(&p->p_smutex);
   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_smutex);
   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_smutex);
   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_smutex);
   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_smutex);
   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 	extern kmutex_t devsw_lock;
   1871 
   1872 	if (newp != NULL || namelen != 0)
   1873 		return (EINVAL);
   1874 
   1875 	start = where = oldp;
   1876 	buflen = *oldlenp;
   1877 	if (where == NULL) {
   1878 		*oldlenp = max_devsw_convs * sizeof kd;
   1879 		return 0;
   1880 	}
   1881 
   1882 	/*
   1883 	 * An array of kinfo_drivers structures
   1884 	 */
   1885 	error = 0;
   1886 	sysctl_unlock();
   1887 	mutex_enter(&devsw_lock);
   1888 	for (i = 0; i < max_devsw_convs; i++) {
   1889 		dname = devsw_conv[i].d_name;
   1890 		if (dname == NULL)
   1891 			continue;
   1892 		if (buflen < sizeof kd) {
   1893 			error = ENOMEM;
   1894 			break;
   1895 		}
   1896 		memset(&kd, 0, sizeof(kd));
   1897 		kd.d_bmajor = devsw_conv[i].d_bmajor;
   1898 		kd.d_cmajor = devsw_conv[i].d_cmajor;
   1899 		strlcpy(kd.d_name, dname, sizeof kd.d_name);
   1900 		mutex_exit(&devsw_lock);
   1901 		error = dcopyout(l, &kd, where, sizeof kd);
   1902 		mutex_enter(&devsw_lock);
   1903 		if (error != 0)
   1904 			break;
   1905 		buflen -= sizeof kd;
   1906 		where += sizeof kd;
   1907 	}
   1908 	mutex_exit(&devsw_lock);
   1909 	sysctl_relock();
   1910 	*oldlenp = where - start;
   1911 	return error;
   1912 }
   1913 
   1914 /*
   1915  * sysctl helper function for kern.file2
   1916  */
   1917 static int
   1918 sysctl_kern_file2(SYSCTLFN_ARGS)
   1919 {
   1920 	struct proc *p;
   1921 	struct file *fp, *tp, *np;
   1922 	struct filedesc *fd;
   1923 	struct kinfo_file kf;
   1924 	char *dp;
   1925 	u_int i, op;
   1926 	size_t len, needed, elem_size, out_size;
   1927 	int error, arg, elem_count;
   1928 	fdfile_t *ff;
   1929 
   1930 	if (namelen == 1 && name[0] == CTL_QUERY)
   1931 		return (sysctl_query(SYSCTLFN_CALL(rnode)));
   1932 
   1933 	if (namelen != 4)
   1934 		return (EINVAL);
   1935 
   1936 	error = 0;
   1937 	dp = oldp;
   1938 	len = (oldp != NULL) ? *oldlenp : 0;
   1939 	op = name[0];
   1940 	arg = name[1];
   1941 	elem_size = name[2];
   1942 	elem_count = name[3];
   1943 	out_size = MIN(sizeof(kf), elem_size);
   1944 	needed = 0;
   1945 
   1946 	if (elem_size < 1 || elem_count < 0)
   1947 		return (EINVAL);
   1948 
   1949 	switch (op) {
   1950 	case KERN_FILE_BYFILE:
   1951 		/*
   1952 		 * doesn't use arg so it must be zero
   1953 		 */
   1954 		if (arg != 0)
   1955 			return (EINVAL);
   1956 		sysctl_unlock();
   1957 		/*
   1958 		 * allocate dummy file descriptor to make position in list
   1959 		 */
   1960 		if ((tp = fgetdummy()) == NULL) {
   1961 		 	sysctl_relock();
   1962 			return ENOMEM;
   1963 		}
   1964 		mutex_enter(&filelist_lock);
   1965 		for (fp = LIST_FIRST(&filehead); fp != NULL; fp = np) {
   1966 			np = LIST_NEXT(fp, f_list);
   1967 			mutex_enter(&fp->f_lock);
   1968 			if (fp->f_count == 0) {
   1969 				mutex_exit(&fp->f_lock);
   1970 				continue;
   1971 			}
   1972 			/*
   1973 			 * XXX Need to prevent that from being an alternative
   1974 			 * XXX way for getting process information.
   1975 			 */
   1976 			if (kauth_authorize_generic(l->l_cred,
   1977 			    KAUTH_GENERIC_CANSEE, fp->f_cred) != 0) {
   1978 				mutex_exit(&fp->f_lock);
   1979 				continue;
   1980 			}
   1981 			if (len >= elem_size && elem_count > 0) {
   1982 				fill_file(&kf, fp, NULL, 0, 0);
   1983 				LIST_INSERT_AFTER(fp, tp, f_list);
   1984 				mutex_exit(&fp->f_lock);
   1985 				mutex_exit(&filelist_lock);
   1986 				error = dcopyout(l, &kf, dp, out_size);
   1987 				mutex_enter(&filelist_lock);
   1988 				np = LIST_NEXT(tp, f_list);
   1989 				LIST_REMOVE(tp, f_list);
   1990 				if (error) {
   1991 					break;
   1992 				}
   1993 				dp += elem_size;
   1994 				len -= elem_size;
   1995 			} else {
   1996 				mutex_exit(&fp->f_lock);
   1997 			}
   1998 			if (elem_count > 0) {
   1999 				needed += elem_size;
   2000 				if (elem_count != INT_MAX)
   2001 					elem_count--;
   2002 			}
   2003 		}
   2004 		mutex_exit(&filelist_lock);
   2005 		fputdummy(tp);
   2006 		sysctl_relock();
   2007 		break;
   2008 	case KERN_FILE_BYPID:
   2009 		if (arg < -1)
   2010 			/* -1 means all processes */
   2011 			return (EINVAL);
   2012 		sysctl_unlock();
   2013 		mutex_enter(&proclist_lock);
   2014 		LIST_FOREACH(p, &allproc, p_list) {
   2015 			if (p->p_stat == SIDL) {
   2016 				/* skip embryonic processes */
   2017 				continue;
   2018 			}
   2019 			if (arg > 0 && p->p_pid != arg) {
   2020 				/* pick only the one we want */
   2021 				/* XXX want 0 to mean "kernel files" */
   2022 				continue;
   2023 			}
   2024 			mutex_enter(&p->p_mutex);
   2025 			error = kauth_authorize_process(l->l_cred,
   2026 			    KAUTH_PROCESS_CANSEE, p,
   2027 			    KAUTH_ARG(KAUTH_REQ_PROCESS_CANSEE_OPENFILES),
   2028 			    NULL, NULL);
   2029 			mutex_exit(&p->p_mutex);
   2030 			if (error != 0) {
   2031 				continue;
   2032 			}
   2033 
   2034 			/*
   2035 			 * Grab a hold on the process.
   2036 			 */
   2037 			if (!rw_tryenter(&p->p_reflock, RW_READER)) {
   2038 				continue;
   2039 			}
   2040 			mutex_exit(&proclist_lock);
   2041 
   2042 			/* XXX Do we need to check permission per file? */
   2043 			fd = p->p_fd;
   2044 			mutex_enter(&fd->fd_lock);
   2045 			for (i = 0; i < fd->fd_nfiles; i++) {
   2046 				if ((ff = fd->fd_ofiles[i]) == NULL) {
   2047 					continue;
   2048 				}
   2049 				mutex_enter(&ff->ff_lock);
   2050 				if ((fp = ff->ff_file) == NULL) {
   2051 					mutex_exit(&ff->ff_lock);
   2052 					continue;
   2053 				}
   2054 				if (len >= elem_size && elem_count > 0) {
   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(&fp->f_lock);
   2067 					mutex_exit(&ff->ff_lock);
   2068 				}
   2069 				if (elem_count > 0) {
   2070 					needed += elem_size;
   2071 					if (elem_count != INT_MAX)
   2072 						elem_count--;
   2073 				}
   2074 			}
   2075 			mutex_exit(&fd->fd_lock);
   2076 
   2077 			/*
   2078 			 * Release reference to process.
   2079 			 */
   2080 			mutex_enter(&proclist_lock);
   2081 			rw_exit(&p->p_reflock);
   2082 		}
   2083 		mutex_exit(&proclist_lock);
   2084 		sysctl_relock();
   2085 		break;
   2086 	default:
   2087 		return (EINVAL);
   2088 	}
   2089 
   2090 	if (oldp == NULL)
   2091 		needed += KERN_FILESLOP * elem_size;
   2092 	*oldlenp = needed;
   2093 
   2094 	return (error);
   2095 }
   2096 
   2097 static void
   2098 fill_file(struct kinfo_file *kp, const file_t *fp, const fdfile_t *ff,
   2099 	  int i, pid_t pid)
   2100 {
   2101 
   2102 	memset(kp, 0, sizeof(*kp));
   2103 
   2104 	kp->ki_fileaddr =	PTRTOUINT64(fp);
   2105 	kp->ki_flag =		fp->f_flag;
   2106 	kp->ki_iflags =		fp->f_iflags;
   2107 	kp->ki_ftype =		fp->f_type;
   2108 	kp->ki_count =		fp->f_count;
   2109 	kp->ki_msgcount =	fp->f_msgcount;
   2110 	kp->ki_fucred =		PTRTOUINT64(fp->f_cred);
   2111 	kp->ki_fuid =		kauth_cred_geteuid(fp->f_cred);
   2112 	kp->ki_fgid =		kauth_cred_getegid(fp->f_cred);
   2113 	kp->ki_fops =		PTRTOUINT64(fp->f_ops);
   2114 	kp->ki_foffset =	fp->f_offset;
   2115 	kp->ki_fdata =		PTRTOUINT64(fp->f_data);
   2116 
   2117 	/* vnode information to glue this file to something */
   2118 	if (fp->f_type == DTYPE_VNODE) {
   2119 		struct vnode *vp = (struct vnode *)fp->f_data;
   2120 
   2121 		kp->ki_vun =	PTRTOUINT64(vp->v_un.vu_socket);
   2122 		kp->ki_vsize =	vp->v_size;
   2123 		kp->ki_vtype =	vp->v_type;
   2124 		kp->ki_vtag =	vp->v_tag;
   2125 		kp->ki_vdata =	PTRTOUINT64(vp->v_data);
   2126 	}
   2127 
   2128 	/* process information when retrieved via KERN_FILE_BYPID */
   2129 	if (ff != NULL) {
   2130 		kp->ki_pid =		pid;
   2131 		kp->ki_fd =		i;
   2132 		kp->ki_ofileflags =	ff->ff_exclose;
   2133 		kp->ki_usecount =	ff->ff_refcnt;
   2134 	}
   2135 }
   2136 
   2137 static int
   2138 sysctl_doeproc(SYSCTLFN_ARGS)
   2139 {
   2140 	struct eproc *eproc;
   2141 	struct kinfo_proc2 *kproc2;
   2142 	struct kinfo_proc *dp;
   2143 	struct proc *p;
   2144 	char *where, *dp2;
   2145 	int type, op, arg;
   2146 	u_int elem_size, elem_count;
   2147 	size_t buflen, needed;
   2148 	bool match;
   2149 	int error;
   2150 
   2151 	if (namelen == 1 && name[0] == CTL_QUERY)
   2152 		return (sysctl_query(SYSCTLFN_CALL(rnode)));
   2153 
   2154 	dp = oldp;
   2155 	dp2 = where = oldp;
   2156 	buflen = where != NULL ? *oldlenp : 0;
   2157 	error = 0;
   2158 	needed = 0;
   2159 	type = rnode->sysctl_num;
   2160 
   2161 	if (type == KERN_PROC) {
   2162 		if (namelen != 2 && !(namelen == 1 && name[0] == KERN_PROC_ALL))
   2163 			return (EINVAL);
   2164 		op = name[0];
   2165 		if (op != KERN_PROC_ALL)
   2166 			arg = name[1];
   2167 		else
   2168 			arg = 0;		/* Quell compiler warning */
   2169 		elem_size = elem_count = 0;	/* Ditto */
   2170 	} else {
   2171 		if (namelen != 4)
   2172 			return (EINVAL);
   2173 		op = name[0];
   2174 		arg = name[1];
   2175 		elem_size = name[2];
   2176 		elem_count = name[3];
   2177 	}
   2178 
   2179 	sysctl_unlock();
   2180 
   2181 	if (type == KERN_PROC) {
   2182 		eproc = kmem_alloc(sizeof(*eproc), KM_SLEEP);
   2183 		kproc2 = NULL;
   2184 	} else {
   2185 		eproc = NULL;
   2186 		kproc2 = kmem_alloc(sizeof(*kproc2), KM_SLEEP);
   2187 	}
   2188 
   2189 	mutex_enter(&proclist_lock);
   2190 	LIST_FOREACH(p, &allproc, p_list) {
   2191 		/*
   2192 		 * Skip embryonic processes.
   2193 		 */
   2194 		if (p->p_stat == SIDL)
   2195 			continue;
   2196 
   2197 		mutex_enter(&p->p_mutex);
   2198 		error = kauth_authorize_process(l->l_cred,
   2199 		    KAUTH_PROCESS_CANSEE, p,
   2200 		    KAUTH_ARG(KAUTH_REQ_PROCESS_CANSEE_ENTRY), NULL, NULL);
   2201 		if (error != 0) {
   2202 			mutex_exit(&p->p_mutex);
   2203 			continue;
   2204 		}
   2205 
   2206 		/*
   2207 		 * TODO - make more efficient (see notes below).
   2208 		 * do by session.
   2209 		 */
   2210 		switch (op) {
   2211 		case KERN_PROC_PID:
   2212 			/* could do this with just a lookup */
   2213 			match = (p->p_pid == (pid_t)arg);
   2214 			break;
   2215 
   2216 		case KERN_PROC_PGRP:
   2217 			/* could do this by traversing pgrp */
   2218 			match = (p->p_pgrp->pg_id == (pid_t)arg);
   2219 			break;
   2220 
   2221 		case KERN_PROC_SESSION:
   2222 			match = (p->p_session->s_sid == (pid_t)arg);
   2223 			break;
   2224 
   2225 		case KERN_PROC_TTY:
   2226 			match = true;
   2227 			if (arg == (int) KERN_PROC_TTY_REVOKE) {
   2228 				if ((p->p_lflag & PL_CONTROLT) == 0 ||
   2229 				    p->p_session->s_ttyp == NULL ||
   2230 				    p->p_session->s_ttyvp != NULL) {
   2231 				    	match = false;
   2232 				}
   2233 			} else if ((p->p_lflag & PL_CONTROLT) == 0 ||
   2234 			    p->p_session->s_ttyp == NULL) {
   2235 				if ((dev_t)arg != KERN_PROC_TTY_NODEV) {
   2236 					match = false;
   2237 				}
   2238 			} else if (p->p_session->s_ttyp->t_dev != (dev_t)arg) {
   2239 				match = false;
   2240 			}
   2241 			break;
   2242 
   2243 		case KERN_PROC_UID:
   2244 			match = (kauth_cred_geteuid(p->p_cred) == (uid_t)arg);
   2245 			break;
   2246 
   2247 		case KERN_PROC_RUID:
   2248 			match = (kauth_cred_getuid(p->p_cred) == (uid_t)arg);
   2249 			break;
   2250 
   2251 		case KERN_PROC_GID:
   2252 			match = (kauth_cred_getegid(p->p_cred) == (uid_t)arg);
   2253 			break;
   2254 
   2255 		case KERN_PROC_RGID:
   2256 			match = (kauth_cred_getgid(p->p_cred) == (uid_t)arg);
   2257 			break;
   2258 
   2259 		case KERN_PROC_ALL:
   2260 			match = true;
   2261 			/* allow everything */
   2262 			break;
   2263 
   2264 		default:
   2265 			error = EINVAL;
   2266 			mutex_exit(&p->p_mutex);
   2267 			goto cleanup;
   2268 		}
   2269 		if (!match) {
   2270 			mutex_exit(&p->p_mutex);
   2271 			continue;
   2272 		}
   2273 
   2274 		/*
   2275 		 * Grab a hold on the process.
   2276 		 */
   2277 		if (!rw_tryenter(&p->p_reflock, RW_READER)) {
   2278 			mutex_exit(&p->p_mutex);
   2279 			continue;
   2280 		}
   2281 
   2282 		if (type == KERN_PROC) {
   2283 			if (buflen >= sizeof(struct kinfo_proc)) {
   2284 				fill_eproc(p, eproc);
   2285 				mutex_exit(&p->p_mutex);
   2286 				mutex_exit(&proclist_lock);
   2287 				error = dcopyout(l, p, &dp->kp_proc,
   2288 				    sizeof(struct proc));
   2289 				mutex_enter(&proclist_lock);
   2290 				if (error) {
   2291 					rw_exit(&p->p_reflock);
   2292 					goto cleanup;
   2293 				}
   2294 				error = dcopyout(l, eproc, &dp->kp_eproc,
   2295 				    sizeof(*eproc));
   2296 				if (error) {
   2297 					rw_exit(&p->p_reflock);
   2298 					goto cleanup;
   2299 				}
   2300 				dp++;
   2301 				buflen -= sizeof(struct kinfo_proc);
   2302 			} else {
   2303 				mutex_exit(&p->p_mutex);
   2304 			}
   2305 			needed += sizeof(struct kinfo_proc);
   2306 		} else { /* KERN_PROC2 */
   2307 			if (buflen >= elem_size && elem_count > 0) {
   2308 				fill_kproc2(p, kproc2);
   2309 				mutex_exit(&p->p_mutex);
   2310 				mutex_exit(&proclist_lock);
   2311 				/*
   2312 				 * Copy out elem_size, but not larger than
   2313 				 * the size of a struct kinfo_proc2.
   2314 				 */
   2315 				error = dcopyout(l, kproc2, dp2,
   2316 				    min(sizeof(*kproc2), elem_size));
   2317 				mutex_enter(&proclist_lock);
   2318 				if (error) {
   2319 					rw_exit(&p->p_reflock);
   2320 					goto cleanup;
   2321 				}
   2322 				dp2 += elem_size;
   2323 				buflen -= elem_size;
   2324 				elem_count--;
   2325 			} else {
   2326 				mutex_exit(&p->p_mutex);
   2327 			}
   2328 			needed += elem_size;
   2329 		}
   2330 
   2331 		/*
   2332 		 * Release reference to process.
   2333 		 */
   2334 		rw_exit(&p->p_reflock);
   2335 	}
   2336 	mutex_exit(&proclist_lock);
   2337 
   2338 	if (where != NULL) {
   2339 		if (type == KERN_PROC)
   2340 			*oldlenp = (char *)dp - where;
   2341 		else
   2342 			*oldlenp = dp2 - where;
   2343 		if (needed > *oldlenp) {
   2344 			error = ENOMEM;
   2345 			goto out;
   2346 		}
   2347 	} else {
   2348 		needed += KERN_PROCSLOP;
   2349 		*oldlenp = needed;
   2350 	}
   2351 	if (kproc2)
   2352 		kmem_free(kproc2, sizeof(*kproc2));
   2353 	if (eproc)
   2354 		kmem_free(eproc, sizeof(*eproc));
   2355 	sysctl_relock();
   2356 	return 0;
   2357  cleanup:
   2358 	mutex_exit(&proclist_lock);
   2359  out:
   2360 	if (kproc2)
   2361 		kmem_free(kproc2, sizeof(*kproc2));
   2362 	if (eproc)
   2363 		kmem_free(eproc, sizeof(*eproc));
   2364 	sysctl_relock();
   2365 	return error;
   2366 }
   2367 
   2368 /*
   2369  * sysctl helper routine for kern.proc_args pseudo-subtree.
   2370  */
   2371 static int
   2372 sysctl_kern_proc_args(SYSCTLFN_ARGS)
   2373 {
   2374 	struct ps_strings pss;
   2375 	struct proc *p;
   2376 	size_t len, i;
   2377 	struct uio auio;
   2378 	struct iovec aiov;
   2379 	pid_t pid;
   2380 	int nargv, type, error, argvlen;
   2381 	char *arg;
   2382 	char **argv = NULL;
   2383 	char *tmp;
   2384 	struct vmspace *vmspace;
   2385 	vaddr_t psstr_addr;
   2386 	vaddr_t offsetn;
   2387 	vaddr_t offsetv;
   2388 
   2389 	if (namelen == 1 && name[0] == CTL_QUERY)
   2390 		return (sysctl_query(SYSCTLFN_CALL(rnode)));
   2391 
   2392 	if (newp != NULL || namelen != 2)
   2393 		return (EINVAL);
   2394 	pid = name[0];
   2395 	type = name[1];
   2396 	argv = NULL;
   2397 	argvlen = 0;
   2398 
   2399 	switch (type) {
   2400 	case KERN_PROC_ARGV:
   2401 	case KERN_PROC_NARGV:
   2402 	case KERN_PROC_ENV:
   2403 	case KERN_PROC_NENV:
   2404 		/* ok */
   2405 		break;
   2406 	default:
   2407 		return (EINVAL);
   2408 	}
   2409 
   2410 	sysctl_unlock();
   2411 
   2412 	/* check pid */
   2413 	mutex_enter(&proclist_lock);
   2414 	if ((p = p_find(pid, PFIND_LOCKED)) == NULL) {
   2415 		error = EINVAL;
   2416 		goto out_locked;
   2417 	}
   2418 	mutex_enter(&p->p_mutex);
   2419 
   2420 	/* Check permission. */
   2421 	if (type == KERN_PROC_ARGV || type == KERN_PROC_NARGV)
   2422 		error = kauth_authorize_process(l->l_cred, KAUTH_PROCESS_CANSEE,
   2423 		    p, KAUTH_ARG(KAUTH_REQ_PROCESS_CANSEE_ARGS), NULL, NULL);
   2424 	else if (type == KERN_PROC_ENV || type == KERN_PROC_NENV)
   2425 		error = kauth_authorize_process(l->l_cred, KAUTH_PROCESS_CANSEE,
   2426 		    p, KAUTH_ARG(KAUTH_REQ_PROCESS_CANSEE_ENV), NULL, NULL);
   2427 	else
   2428 		error = EINVAL; /* XXXGCC */
   2429 	if (error) {
   2430 		mutex_exit(&p->p_mutex);
   2431 		goto out_locked;
   2432 	}
   2433 
   2434 	if (oldp == NULL) {
   2435 		if (type == KERN_PROC_NARGV || type == KERN_PROC_NENV)
   2436 			*oldlenp = sizeof (int);
   2437 		else
   2438 			*oldlenp = ARG_MAX;	/* XXX XXX XXX */
   2439 		error = 0;
   2440 		mutex_exit(&p->p_mutex);
   2441 		goto out_locked;
   2442 	}
   2443 
   2444 	/*
   2445 	 * Zombies don't have a stack, so we can't read their psstrings.
   2446 	 * System processes also don't have a user stack.
   2447 	 */
   2448 	if (P_ZOMBIE(p) || (p->p_flag & PK_SYSTEM) != 0) {
   2449 		error = EINVAL;
   2450 		mutex_exit(&p->p_mutex);
   2451 		goto out_locked;
   2452 	}
   2453 
   2454 	/*
   2455 	 * Lock the process down in memory.
   2456 	 */
   2457 	psstr_addr = (vaddr_t)p->p_psstr;
   2458 	if (type == KERN_PROC_ARGV || type == KERN_PROC_NARGV) {
   2459 		offsetn = p->p_psnargv;
   2460 		offsetv = p->p_psargv;
   2461 	} else {
   2462 		offsetn = p->p_psnenv;
   2463 		offsetv = p->p_psenv;
   2464 	}
   2465 	vmspace = p->p_vmspace;
   2466 	uvmspace_addref(vmspace);
   2467 	mutex_exit(&p->p_mutex);
   2468 	mutex_exit(&proclist_lock);
   2469 
   2470 	/*
   2471 	 * Allocate a temporary buffer to hold the arguments.
   2472 	 */
   2473 	arg = kmem_alloc(PAGE_SIZE, KM_SLEEP);
   2474 
   2475 	/*
   2476 	 * Read in the ps_strings structure.
   2477 	 */
   2478 	aiov.iov_base = &pss;
   2479 	aiov.iov_len = sizeof(pss);
   2480 	auio.uio_iov = &aiov;
   2481 	auio.uio_iovcnt = 1;
   2482 	auio.uio_offset = psstr_addr;
   2483 	auio.uio_resid = sizeof(pss);
   2484 	auio.uio_rw = UIO_READ;
   2485 	UIO_SETUP_SYSSPACE(&auio);
   2486 	error = uvm_io(&vmspace->vm_map, &auio);
   2487 	if (error)
   2488 		goto done;
   2489 
   2490 	memcpy(&nargv, (char *)&pss + offsetn, sizeof(nargv));
   2491 	if (type == KERN_PROC_NARGV || type == KERN_PROC_NENV) {
   2492 		error = dcopyout(l, &nargv, oldp, sizeof(nargv));
   2493 		*oldlenp = sizeof(nargv);
   2494 		goto done;
   2495 	}
   2496 	/*
   2497 	 * Now read the address of the argument vector.
   2498 	 */
   2499 	switch (type) {
   2500 	case KERN_PROC_ARGV:
   2501 		/* FALLTHROUGH */
   2502 	case KERN_PROC_ENV:
   2503 		memcpy(&tmp, (char *)&pss + offsetv, sizeof(tmp));
   2504 		break;
   2505 	default:
   2506 		error = EINVAL;
   2507 		goto done;
   2508 	}
   2509 
   2510 #ifdef COMPAT_NETBSD32
   2511 	if (p->p_flag & PK_32)
   2512 		len = sizeof(netbsd32_charp) * nargv;
   2513 	else
   2514 #endif
   2515 		len = sizeof(char *) * nargv;
   2516 
   2517 	if ((argvlen = len) != 0)
   2518 		argv = kmem_alloc(len, KM_SLEEP);
   2519 
   2520 	aiov.iov_base = argv;
   2521 	aiov.iov_len = len;
   2522 	auio.uio_iov = &aiov;
   2523 	auio.uio_iovcnt = 1;
   2524 	auio.uio_offset = (off_t)(unsigned long)tmp;
   2525 	auio.uio_resid = len;
   2526 	auio.uio_rw = UIO_READ;
   2527 	UIO_SETUP_SYSSPACE(&auio);
   2528 	error = uvm_io(&vmspace->vm_map, &auio);
   2529 	if (error)
   2530 		goto done;
   2531 
   2532 	/*
   2533 	 * Now copy each string.
   2534 	 */
   2535 	len = 0; /* bytes written to user buffer */
   2536 	for (i = 0; i < nargv; i++) {
   2537 		int finished = 0;
   2538 		vaddr_t base;
   2539 		size_t xlen;
   2540 		int j;
   2541 
   2542 #ifdef COMPAT_NETBSD32
   2543 		if (p->p_flag & PK_32) {
   2544 			netbsd32_charp *argv32;
   2545 
   2546 			argv32 = (netbsd32_charp *)argv;
   2547 			base = (vaddr_t)NETBSD32PTR64(argv32[i]);
   2548 		} else
   2549 #endif
   2550 			base = (vaddr_t)argv[i];
   2551 
   2552 		/*
   2553 		 * The program has messed around with its arguments,
   2554 		 * possibly deleting some, and replacing them with
   2555 		 * NULL's. Treat this as the last argument and not
   2556 		 * a failure.
   2557 		 */
   2558 		if (base == 0)
   2559 			break;
   2560 
   2561 		while (!finished) {
   2562 			xlen = PAGE_SIZE - (base & PAGE_MASK);
   2563 
   2564 			aiov.iov_base = arg;
   2565 			aiov.iov_len = PAGE_SIZE;
   2566 			auio.uio_iov = &aiov;
   2567 			auio.uio_iovcnt = 1;
   2568 			auio.uio_offset = base;
   2569 			auio.uio_resid = xlen;
   2570 			auio.uio_rw = UIO_READ;
   2571 			UIO_SETUP_SYSSPACE(&auio);
   2572 			error = uvm_io(&vmspace->vm_map, &auio);
   2573 			if (error)
   2574 				goto done;
   2575 
   2576 			/* Look for the end of the string */
   2577 			for (j = 0; j < xlen; j++) {
   2578 				if (arg[j] == '\0') {
   2579 					xlen = j + 1;
   2580 					finished = 1;
   2581 					break;
   2582 				}
   2583 			}
   2584 
   2585 			/* Check for user buffer overflow */
   2586 			if (len + xlen > *oldlenp) {
   2587 				finished = 1;
   2588 				if (len > *oldlenp)
   2589 					xlen = 0;
   2590 				else
   2591 					xlen = *oldlenp - len;
   2592 			}
   2593 
   2594 			/* Copyout the page */
   2595 			error = dcopyout(l, arg, (char *)oldp + len, xlen);
   2596 			if (error)
   2597 				goto done;
   2598 
   2599 			len += xlen;
   2600 			base += xlen;
   2601 		}
   2602 	}
   2603 	*oldlenp = len;
   2604 
   2605 done:
   2606 	if (argvlen != 0)
   2607 		kmem_free(argv, argvlen);
   2608 	uvmspace_free(vmspace);
   2609 	kmem_free(arg, PAGE_SIZE);
   2610 	sysctl_relock();
   2611 	return error;
   2612 
   2613 out_locked:
   2614 	mutex_exit(&proclist_lock);
   2615 	sysctl_relock();
   2616 	return error;
   2617 }
   2618 
   2619 static int
   2620 sysctl_security_setidcore(SYSCTLFN_ARGS)
   2621 {
   2622 	int newsize, error;
   2623 	struct sysctlnode node;
   2624 
   2625 	node = *rnode;
   2626 	node.sysctl_data = &newsize;
   2627 	newsize = *(int *)rnode->sysctl_data;
   2628 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
   2629 	if (error || newp == NULL)
   2630 		return error;
   2631 
   2632 	if (kauth_authorize_system(l->l_cred, KAUTH_SYSTEM_SETIDCORE,
   2633 	    0, NULL, NULL, NULL))
   2634 		return (EPERM);
   2635 
   2636 	*(int *)rnode->sysctl_data = newsize;
   2637 
   2638 	return 0;
   2639 }
   2640 
   2641 static int
   2642 sysctl_security_setidcorename(SYSCTLFN_ARGS)
   2643 {
   2644 	int error;
   2645 	char *newsetidcorename;
   2646 	struct sysctlnode node;
   2647 
   2648 	newsetidcorename = PNBUF_GET();
   2649 	node = *rnode;
   2650 	node.sysctl_data = newsetidcorename;
   2651 	memcpy(node.sysctl_data, rnode->sysctl_data, MAXPATHLEN);
   2652 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
   2653 	if (error || newp == NULL) {
   2654 		goto out;
   2655 	}
   2656 	if (kauth_authorize_system(l->l_cred, KAUTH_SYSTEM_SETIDCORE,
   2657 	    0, NULL, NULL, NULL)) {
   2658 		error = EPERM;
   2659 		goto out;
   2660 	}
   2661 	if (strlen(newsetidcorename) == 0) {
   2662 		error = EINVAL;
   2663 		goto out;
   2664 	}
   2665 	memcpy(rnode->sysctl_data, node.sysctl_data, MAXPATHLEN);
   2666 out:
   2667 	PNBUF_PUT(newsetidcorename);
   2668 	return error;
   2669 }
   2670 
   2671 /*
   2672  * sysctl helper routine for kern.cp_id node. Maps cpus to their
   2673  * cpuids.
   2674  */
   2675 static int
   2676 sysctl_kern_cpid(SYSCTLFN_ARGS)
   2677 {
   2678 	struct sysctlnode node = *rnode;
   2679 	uint64_t *cp_id = NULL;
   2680 	int error, n = ncpu;
   2681 	struct cpu_info *ci;
   2682 	CPU_INFO_ITERATOR cii;
   2683 
   2684 	/*
   2685 	 * Here you may either retrieve a single cpu id or the whole
   2686 	 * set. The size you get back when probing depends on what
   2687 	 * you ask for.
   2688 	 */
   2689 	switch (namelen) {
   2690 	case 0:
   2691 		node.sysctl_size = n * sizeof(uint64_t);
   2692 		n = -2; /* ALL */
   2693 		break;
   2694 	case 1:
   2695 		if (name[0] < 0 || name[0] >= n)
   2696 			return (ENOENT); /* ENOSUCHPROCESSOR */
   2697 		node.sysctl_size = sizeof(uint64_t);
   2698 		n = name[0];
   2699 		/*
   2700 		 * adjust these so that sysctl_lookup() will be happy
   2701 		 */
   2702 		name++;
   2703 		namelen--;
   2704 		break;
   2705 	default:
   2706 		return (EINVAL);
   2707 	}
   2708 
   2709 	cp_id = kmem_alloc(node.sysctl_size, KM_SLEEP);
   2710 	if (cp_id == NULL)
   2711 		return (ENOMEM);
   2712 	node.sysctl_data = cp_id;
   2713 	memset(cp_id, 0, node.sysctl_size);
   2714 
   2715 	for (CPU_INFO_FOREACH(cii, ci)) {
   2716 		if (n <= 0)
   2717 			cp_id[0] = ci->ci_cpuid;
   2718 		/*
   2719 		 * if a specific processor was requested and we just
   2720 		 * did it, we're done here
   2721 		 */
   2722 		if (n == 0)
   2723 			break;
   2724 		/*
   2725 		 * if doing "all", skip to next cp_id slot for next processor
   2726 		 */
   2727 		if (n == -2)
   2728 			cp_id++;
   2729 		/*
   2730 		 * if we're doing a specific processor, we're one
   2731 		 * processor closer
   2732 		 */
   2733 		if (n > 0)
   2734 			n--;
   2735 	}
   2736 
   2737 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
   2738 	kmem_free(node.sysctl_data, node.sysctl_size);
   2739 	return (error);
   2740 }
   2741 
   2742 /*
   2743  * sysctl helper routine for hw.usermem and hw.usermem64. Values are
   2744  * calculate on the fly taking into account integer overflow and the
   2745  * current wired count.
   2746  */
   2747 static int
   2748 sysctl_hw_usermem(SYSCTLFN_ARGS)
   2749 {
   2750 	u_int ui;
   2751 	u_quad_t uq;
   2752 	struct sysctlnode node;
   2753 
   2754 	node = *rnode;
   2755 	switch (rnode->sysctl_num) {
   2756 	    case HW_USERMEM:
   2757 		if ((ui = physmem - uvmexp.wired) > (UINT_MAX / PAGE_SIZE))
   2758 			ui = UINT_MAX;
   2759 		else
   2760 			ui *= PAGE_SIZE;
   2761 		node.sysctl_data = &ui;
   2762 		break;
   2763 	case HW_USERMEM64:
   2764 		uq = (u_quad_t)(physmem - uvmexp.wired) * PAGE_SIZE;
   2765 		node.sysctl_data = &uq;
   2766 		break;
   2767 	default:
   2768 		return (EINVAL);
   2769 	}
   2770 
   2771 	return (sysctl_lookup(SYSCTLFN_CALL(&node)));
   2772 }
   2773 
   2774 /*
   2775  * sysctl helper routine for kern.cnmagic node. Pulls the old value
   2776  * out, encoded, and stuffs the new value in for decoding.
   2777  */
   2778 static int
   2779 sysctl_hw_cnmagic(SYSCTLFN_ARGS)
   2780 {
   2781 	char magic[CNS_LEN];
   2782 	int error;
   2783 	struct sysctlnode node;
   2784 
   2785 	if (oldp)
   2786 		cn_get_magic(magic, CNS_LEN);
   2787 	node = *rnode;
   2788 	node.sysctl_data = &magic[0];
   2789 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
   2790 	if (error || newp == NULL)
   2791 		return (error);
   2792 
   2793 	return (cn_set_magic(magic));
   2794 }
   2795 
   2796 /*
   2797  * ********************************************************************
   2798  * section 3: public helper routines that are used for more than one
   2799  * node
   2800  * ********************************************************************
   2801  */
   2802 
   2803 /*
   2804  * sysctl helper routine for the kern.root_device node and some ports'
   2805  * machdep.root_device nodes.
   2806  */
   2807 int
   2808 sysctl_root_device(SYSCTLFN_ARGS)
   2809 {
   2810 	struct sysctlnode node;
   2811 
   2812 	node = *rnode;
   2813 	node.sysctl_data = root_device->dv_xname;
   2814 	node.sysctl_size = strlen(root_device->dv_xname) + 1;
   2815 	return (sysctl_lookup(SYSCTLFN_CALL(&node)));
   2816 }
   2817 
   2818 /*
   2819  * sysctl helper routine for kern.consdev, dependent on the current
   2820  * state of the console. Also used for machdep.console_device on some
   2821  * ports.
   2822  */
   2823 int
   2824 sysctl_consdev(SYSCTLFN_ARGS)
   2825 {
   2826 	dev_t consdev;
   2827 	struct sysctlnode node;
   2828 
   2829 	if (cn_tab != NULL)
   2830 		consdev = cn_tab->cn_dev;
   2831 	else
   2832 		consdev = NODEV;
   2833 	node = *rnode;
   2834 	node.sysctl_data = &consdev;
   2835 	node.sysctl_size = sizeof(consdev);
   2836 	return (sysctl_lookup(SYSCTLFN_CALL(&node)));
   2837 }
   2838 
   2839 /*
   2840  * ********************************************************************
   2841  * section 4: support for some helpers
   2842  * ********************************************************************
   2843  */
   2844 
   2845 /*
   2846  * Fill in a kinfo_proc2 structure for the specified process.
   2847  */
   2848 static void
   2849 fill_kproc2(struct proc *p, struct kinfo_proc2 *ki)
   2850 {
   2851 	struct tty *tp;
   2852 	struct lwp *l, *l2;
   2853 	struct timeval ut, st, rt;
   2854 	sigset_t ss1, ss2;
   2855 	struct rusage ru;
   2856 
   2857 	KASSERT(mutex_owned(&proclist_lock));
   2858 	KASSERT(mutex_owned(&p->p_mutex));
   2859 
   2860 	memset(ki, 0, sizeof(*ki));
   2861 
   2862 	ki->p_paddr = PTRTOUINT64(p);
   2863 	ki->p_fd = PTRTOUINT64(p->p_fd);
   2864 	ki->p_cwdi = PTRTOUINT64(p->p_cwdi);
   2865 	ki->p_stats = PTRTOUINT64(p->p_stats);
   2866 	ki->p_limit = PTRTOUINT64(p->p_limit);
   2867 	ki->p_vmspace = PTRTOUINT64(p->p_vmspace);
   2868 	ki->p_sigacts = PTRTOUINT64(p->p_sigacts);
   2869 	ki->p_sess = PTRTOUINT64(p->p_session);
   2870 	ki->p_tsess = 0;	/* may be changed if controlling tty below */
   2871 	ki->p_ru = PTRTOUINT64(&p->p_stats->p_ru);
   2872 
   2873 	ki->p_eflag = 0;
   2874 	ki->p_exitsig = p->p_exitsig;
   2875 
   2876 	ki->p_flag = sysctl_map_flags(sysctl_flagmap, p->p_flag);
   2877 	ki->p_flag |= sysctl_map_flags(sysctl_sflagmap, p->p_sflag);
   2878 	ki->p_flag |= sysctl_map_flags(sysctl_slflagmap, p->p_slflag);
   2879 	ki->p_flag |= sysctl_map_flags(sysctl_lflagmap, p->p_lflag);
   2880 	ki->p_flag |= sysctl_map_flags(sysctl_stflagmap, p->p_stflag);
   2881 
   2882 	ki->p_pid = p->p_pid;
   2883 	if (p->p_pptr)
   2884 		ki->p_ppid = p->p_pptr->p_pid;
   2885 	else
   2886 		ki->p_ppid = 0;
   2887 	ki->p_sid = p->p_session->s_sid;
   2888 	ki->p__pgid = p->p_pgrp->pg_id;
   2889 
   2890 	ki->p_tpgid = NO_PGID;	/* may be changed if controlling tty below */
   2891 
   2892 	ki->p_uid = kauth_cred_geteuid(p->p_cred);
   2893 	ki->p_ruid = kauth_cred_getuid(p->p_cred);
   2894 	ki->p_gid = kauth_cred_getegid(p->p_cred);
   2895 	ki->p_rgid = kauth_cred_getgid(p->p_cred);
   2896 	ki->p_svuid = kauth_cred_getsvuid(p->p_cred);
   2897 	ki->p_svgid = kauth_cred_getsvgid(p->p_cred);
   2898 
   2899 	ki->p_ngroups = kauth_cred_ngroups(p->p_cred);
   2900 	kauth_cred_getgroups(p->p_cred, ki->p_groups,
   2901 	    min(ki->p_ngroups, sizeof(ki->p_groups) / sizeof(ki->p_groups[0])),
   2902 	    UIO_SYSSPACE);
   2903 
   2904 	ki->p_jobc = p->p_pgrp->pg_jobc;
   2905 	if ((p->p_lflag & PL_CONTROLT) && (tp = p->p_session->s_ttyp)) {
   2906 		ki->p_tdev = tp->t_dev;
   2907 		ki->p_tpgid = tp->t_pgrp ? tp->t_pgrp->pg_id : NO_PGID;
   2908 		ki->p_tsess = PTRTOUINT64(tp->t_session);
   2909 	} else {
   2910 		ki->p_tdev = NODEV;
   2911 	}
   2912 
   2913 	mutex_enter(&p->p_smutex);
   2914 
   2915 	ki->p_uticks = p->p_uticks;
   2916 	ki->p_sticks = p->p_sticks;
   2917 	ki->p_iticks = p->p_iticks;
   2918 
   2919 	ki->p_tracep = PTRTOUINT64(p->p_tracep);
   2920 	ki->p_traceflag = p->p_traceflag;
   2921 
   2922 	memcpy(&ki->p_sigignore, &p->p_sigctx.ps_sigignore,sizeof(ki_sigset_t));
   2923 	memcpy(&ki->p_sigcatch, &p->p_sigctx.ps_sigcatch, sizeof(ki_sigset_t));
   2924 
   2925 	ki->p_cpticks = 0;
   2926 	ki->p_pctcpu = p->p_pctcpu;
   2927 	ki->p_estcpu = 0;
   2928 	ss1 = p->p_sigpend.sp_set;
   2929 	LIST_FOREACH(l, &p->p_lwps, l_sibling) {
   2930 		/* This is hardly correct, but... */
   2931 		sigplusset(&l->l_sigpend.sp_set, &ss1);
   2932 		sigplusset(&l->l_sigmask, &ss2);
   2933 		ki->p_cpticks += l->l_cpticks;
   2934 		ki->p_pctcpu += l->l_pctcpu;
   2935 		ki->p_estcpu += l->l_estcpu;
   2936 	}
   2937 	memcpy(&ki->p_siglist, &ss1, sizeof(ki_sigset_t));
   2938 	memcpy(&ki->p_sigmask, &ss2, sizeof(ki_sigset_t));
   2939 
   2940 	ki->p_stat = p->p_stat; /* Will likely be overridden by LWP status */
   2941 	ki->p_realstat = p->p_stat;
   2942 	ki->p_nice = p->p_nice;
   2943 
   2944 	ki->p_xstat = p->p_xstat;
   2945 	ki->p_acflag = p->p_acflag;
   2946 
   2947 	strncpy(ki->p_comm, p->p_comm,
   2948 	    min(sizeof(ki->p_comm), sizeof(p->p_comm)));
   2949 
   2950 	strncpy(ki->p_login, p->p_session->s_login,
   2951 	    min(sizeof ki->p_login - 1, sizeof p->p_session->s_login));
   2952 
   2953 	ki->p_nlwps = p->p_nlwps;
   2954 	ki->p_realflag = ki->p_flag;
   2955 
   2956 	if (p->p_stat == SIDL || P_ZOMBIE(p)) {
   2957 		ki->p_vm_rssize = 0;
   2958 		ki->p_vm_tsize = 0;
   2959 		ki->p_vm_dsize = 0;
   2960 		ki->p_vm_ssize = 0;
   2961 		ki->p_nrlwps = 0;
   2962 		l = NULL;
   2963 	} else {
   2964 		struct vmspace *vm = p->p_vmspace;
   2965 		int tmp;
   2966 
   2967 		ki->p_vm_rssize = vm_resident_count(vm);
   2968 		ki->p_vm_tsize = vm->vm_tsize;
   2969 		ki->p_vm_dsize = vm->vm_dsize;
   2970 		ki->p_vm_ssize = vm->vm_ssize;
   2971 
   2972 		/* Pick a "representative" LWP */
   2973 		l = proc_representative_lwp(p, &tmp, 1);
   2974 		lwp_lock(l);
   2975 		ki->p_nrlwps = tmp;
   2976 		ki->p_forw = 0;
   2977 		ki->p_back = 0;
   2978 		ki->p_addr = PTRTOUINT64(l->l_addr);
   2979 		ki->p_stat = l->l_stat;
   2980 		ki->p_flag |= sysctl_map_flags(sysctl_lwpflagmap, l->l_flag);
   2981 		ki->p_swtime = l->l_swtime;
   2982 		ki->p_slptime = l->l_slptime;
   2983 		if (l->l_stat == LSONPROC)
   2984 			ki->p_schedflags = l->l_cpu->ci_schedstate.spc_flags;
   2985 		else
   2986 			ki->p_schedflags = 0;
   2987 		ki->p_holdcnt = l->l_holdcnt;
   2988 		ki->p_priority = lwp_eprio(l);
   2989 		ki->p_usrpri = l->l_priority;
   2990 		if (l->l_wmesg)
   2991 			strncpy(ki->p_wmesg, l->l_wmesg, sizeof(ki->p_wmesg));
   2992 		ki->p_wchan = PTRTOUINT64(l->l_wchan);
   2993 		lwp_unlock(l);
   2994 	}
   2995 	if (p->p_session->s_ttyvp)
   2996 		ki->p_eflag |= EPROC_CTTY;
   2997 	if (SESS_LEADER(p))
   2998 		ki->p_eflag |= EPROC_SLEADER;
   2999 
   3000 	/* XXX Is this double check necessary? */
   3001 	if (P_ZOMBIE(p)) {
   3002 		ki->p_uvalid = 0;
   3003 		ki->p_rtime_sec = 0;
   3004 		ki->p_rtime_usec = 0;
   3005 	} else {
   3006 		ki->p_uvalid = 1;
   3007 
   3008 		ki->p_ustart_sec = p->p_stats->p_start.tv_sec;
   3009 		ki->p_ustart_usec = p->p_stats->p_start.tv_usec;
   3010 
   3011 		calcru(p, &ut, &st, NULL, &rt);
   3012 		ki->p_rtime_sec = rt.tv_sec;
   3013 		ki->p_rtime_usec = rt.tv_usec;
   3014 		ki->p_uutime_sec = ut.tv_sec;
   3015 		ki->p_uutime_usec = ut.tv_usec;
   3016 		ki->p_ustime_sec = st.tv_sec;
   3017 		ki->p_ustime_usec = st.tv_usec;
   3018 
   3019 		memcpy(&ru, &p->p_stats->p_ru, sizeof(ru));
   3020 		ki->p_uru_nvcsw = 0;
   3021 		ki->p_uru_nivcsw = 0;
   3022 		LIST_FOREACH(l2, &p->p_lwps, l_sibling) {
   3023 			ki->p_uru_nvcsw += (l->l_ncsw - l->l_nivcsw);
   3024 			ki->p_uru_nivcsw += l->l_nivcsw;
   3025 			ruadd(&ru, &l2->l_ru);
   3026 		}
   3027 		ki->p_uru_maxrss = ru.ru_maxrss;
   3028 		ki->p_uru_ixrss = ru.ru_ixrss;
   3029 		ki->p_uru_idrss = ru.ru_idrss;
   3030 		ki->p_uru_isrss = ru.ru_isrss;
   3031 		ki->p_uru_minflt = ru.ru_minflt;
   3032 		ki->p_uru_majflt = ru.ru_majflt;
   3033 		ki->p_uru_nswap = ru.ru_nswap;
   3034 		ki->p_uru_inblock = ru.ru_inblock;
   3035 		ki->p_uru_oublock = ru.ru_oublock;
   3036 		ki->p_uru_msgsnd = ru.ru_msgsnd;
   3037 		ki->p_uru_msgrcv = ru.ru_msgrcv;
   3038 		ki->p_uru_nsignals = ru.ru_nsignals;
   3039 
   3040 		timeradd(&p->p_stats->p_cru.ru_utime,
   3041 			 &p->p_stats->p_cru.ru_stime, &ut);
   3042 		ki->p_uctime_sec = ut.tv_sec;
   3043 		ki->p_uctime_usec = ut.tv_usec;
   3044 	}
   3045 	if (l != NULL)
   3046 		ki->p_cpuid = l->l_cpu->ci_cpuid;
   3047 
   3048 	mutex_exit(&p->p_smutex);
   3049 }
   3050 
   3051 /*
   3052  * Fill in a kinfo_lwp structure for the specified lwp.
   3053  */
   3054 static void
   3055 fill_lwp(struct lwp *l, struct kinfo_lwp *kl)
   3056 {
   3057 	struct proc *p = l->l_proc;
   3058 	struct timeval tv;
   3059 
   3060 	kl->l_forw = 0;
   3061 	kl->l_back = 0;
   3062 	kl->l_laddr = PTRTOUINT64(l);
   3063 	kl->l_addr = PTRTOUINT64(l->l_addr);
   3064 	kl->l_stat = l->l_stat;
   3065 	kl->l_lid = l->l_lid;
   3066 	kl->l_flag = sysctl_map_flags(sysctl_lwpprflagmap, l->l_prflag);
   3067 
   3068 	kl->l_swtime = l->l_swtime;
   3069 	kl->l_slptime = l->l_slptime;
   3070 	if (l->l_stat == LSONPROC)
   3071 		kl->l_schedflags = l->l_cpu->ci_schedstate.spc_flags;
   3072 	else
   3073 		kl->l_schedflags = 0;
   3074 	kl->l_holdcnt = l->l_holdcnt;
   3075 	kl->l_priority = lwp_eprio(l);
   3076 	kl->l_usrpri = l->l_priority;
   3077 	if (l->l_wmesg)
   3078 		strncpy(kl->l_wmesg, l->l_wmesg, sizeof(kl->l_wmesg));
   3079 	kl->l_wchan = PTRTOUINT64(l->l_wchan);
   3080 	kl->l_cpuid = l->l_cpu->ci_cpuid;
   3081 	bintime2timeval(&l->l_rtime, &tv);
   3082 	kl->l_rtime_sec = tv.tv_sec;
   3083 	kl->l_rtime_usec = tv.tv_usec;
   3084 	kl->l_cpticks = l->l_cpticks;
   3085 	kl->l_pctcpu = l->l_pctcpu;
   3086 	kl->l_pid = p->p_pid;
   3087 	if (l->l_name == NULL)
   3088 		kl->l_name[0] = '\0';
   3089 	else
   3090 		strlcpy(kl->l_name, l->l_name, sizeof(kl->l_name));
   3091 }
   3092 
   3093 /*
   3094  * Fill in an eproc structure for the specified process.
   3095  */
   3096 void
   3097 fill_eproc(struct proc *p, struct eproc *ep)
   3098 {
   3099 	struct tty *tp;
   3100 	struct lwp *l;
   3101 
   3102 	KASSERT(mutex_owned(&proclist_lock));
   3103 	KASSERT(mutex_owned(&p->p_mutex));
   3104 
   3105 	ep->e_paddr = p;
   3106 	ep->e_sess = p->p_session;
   3107 	kauth_cred_topcred(p->p_cred, &ep->e_pcred);
   3108 	kauth_cred_toucred(p->p_cred, &ep->e_ucred);
   3109 	if (p->p_stat == SIDL || P_ZOMBIE(p)) {
   3110 		ep->e_vm.vm_rssize = 0;
   3111 		ep->e_vm.vm_tsize = 0;
   3112 		ep->e_vm.vm_dsize = 0;
   3113 		ep->e_vm.vm_ssize = 0;
   3114 		/* ep->e_vm.vm_pmap = XXX; */
   3115 	} else {
   3116 		struct vmspace *vm = p->p_vmspace;
   3117 
   3118 		ep->e_vm.vm_rssize = vm_resident_count(vm);
   3119 		ep->e_vm.vm_tsize = vm->vm_tsize;
   3120 		ep->e_vm.vm_dsize = vm->vm_dsize;
   3121 		ep->e_vm.vm_ssize = vm->vm_ssize;
   3122 
   3123 		/* Pick a "representative" LWP */
   3124 		mutex_enter(&p->p_smutex);
   3125 		l = proc_representative_lwp(p, NULL, 1);
   3126 		lwp_lock(l);
   3127 		if (l->l_wmesg)
   3128 			strncpy(ep->e_wmesg, l->l_wmesg, WMESGLEN);
   3129 		lwp_unlock(l);
   3130 		mutex_exit(&p->p_smutex);
   3131 	}
   3132 	if (p->p_pptr)
   3133 		ep->e_ppid = p->p_pptr->p_pid;
   3134 	else
   3135 		ep->e_ppid = 0;
   3136 	ep->e_pgid = p->p_pgrp->pg_id;
   3137 	ep->e_sid = ep->e_sess->s_sid;
   3138 	ep->e_jobc = p->p_pgrp->pg_jobc;
   3139 	if ((p->p_lflag & PL_CONTROLT) &&
   3140 	    (tp = ep->e_sess->s_ttyp)) {
   3141 		ep->e_tdev = tp->t_dev;
   3142 		ep->e_tpgid = tp->t_pgrp ? tp->t_pgrp->pg_id : NO_PGID;
   3143 		ep->e_tsess = tp->t_session;
   3144 	} else
   3145 		ep->e_tdev = NODEV;
   3146 
   3147 	ep->e_xsize = ep->e_xrssize = 0;
   3148 	ep->e_xccount = ep->e_xswrss = 0;
   3149 	ep->e_flag = ep->e_sess->s_ttyvp ? EPROC_CTTY : 0;
   3150 	if (SESS_LEADER(p))
   3151 		ep->e_flag |= EPROC_SLEADER;
   3152 	strncpy(ep->e_login, ep->e_sess->s_login, MAXLOGNAME);
   3153 }
   3154 
   3155 u_int
   3156 sysctl_map_flags(const u_int *map, u_int word)
   3157 {
   3158 	u_int rv;
   3159 
   3160 	for (rv = 0; *map != 0; map += 2)
   3161 		if ((word & map[0]) != 0)
   3162 			rv |= map[1];
   3163 
   3164 	return rv;
   3165 }
   3166