Home | History | Annotate | Line # | Download | only in kern
init_sysctl.c revision 1.122
      1 /*	$NetBSD: init_sysctl.c,v 1.122 2008/01/30 00:43:47 ad Exp $ */
      2 
      3 /*-
      4  * Copyright (c) 2003, 2007, 2008 The NetBSD Foundation, Inc.
      5  * All rights reserved.
      6  *
      7  * This code is derived from software contributed to The NetBSD Foundation
      8  * by Andrew Brown, and by Andrew Doran.
      9  *
     10  * Redistribution and use in source and binary forms, with or without
     11  * modification, are permitted provided that the following conditions
     12  * are met:
     13  * 1. Redistributions of source code must retain the above copyright
     14  *    notice, this list of conditions and the following disclaimer.
     15  * 2. Redistributions in binary form must reproduce the above copyright
     16  *    notice, this list of conditions and the following disclaimer in the
     17  *    documentation and/or other materials provided with the distribution.
     18  * 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 1.122 2008/01/30 00:43:47 ad 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(l, kaddr, uaddr, len)
    151 	struct lwp *l;
    152 	const void *kaddr;
    153 	void *uaddr;
    154 	size_t len;
    155 {
    156 	int error;
    157 
    158 	error = copyout(kaddr, uaddr, len);
    159 	ktrmibio(-1, UIO_READ, uaddr, len, error);
    160 
    161 	return error;
    162 }
    163 
    164 #ifdef DIAGNOSTIC
    165 static int sysctl_kern_trigger_panic(SYSCTLFN_PROTO);
    166 #endif
    167 static int sysctl_kern_maxvnodes(SYSCTLFN_PROTO);
    168 static int sysctl_kern_rtc_offset(SYSCTLFN_PROTO);
    169 static int sysctl_kern_maxproc(SYSCTLFN_PROTO);
    170 static int sysctl_kern_hostid(SYSCTLFN_PROTO);
    171 static int sysctl_setlen(SYSCTLFN_PROTO);
    172 static int sysctl_kern_clockrate(SYSCTLFN_PROTO);
    173 static int sysctl_kern_file(SYSCTLFN_PROTO);
    174 static int sysctl_msgbuf(SYSCTLFN_PROTO);
    175 static int sysctl_kern_defcorename(SYSCTLFN_PROTO);
    176 static int sysctl_kern_cptime(SYSCTLFN_PROTO);
    177 #if NPTY > 0
    178 static int sysctl_kern_maxptys(SYSCTLFN_PROTO);
    179 #endif /* NPTY > 0 */
    180 static int sysctl_kern_sbmax(SYSCTLFN_PROTO);
    181 static int sysctl_kern_urnd(SYSCTLFN_PROTO);
    182 static int sysctl_kern_arnd(SYSCTLFN_PROTO);
    183 static int sysctl_kern_lwp(SYSCTLFN_PROTO);
    184 static int sysctl_kern_forkfsleep(SYSCTLFN_PROTO);
    185 static int sysctl_kern_root_partition(SYSCTLFN_PROTO);
    186 static int sysctl_kern_drivers(SYSCTLFN_PROTO);
    187 static int sysctl_kern_file2(SYSCTLFN_PROTO);
    188 static int sysctl_security_setidcore(SYSCTLFN_PROTO);
    189 static int sysctl_security_setidcorename(SYSCTLFN_PROTO);
    190 static int sysctl_kern_cpid(SYSCTLFN_PROTO);
    191 static int sysctl_doeproc(SYSCTLFN_PROTO);
    192 static int sysctl_kern_proc_args(SYSCTLFN_PROTO);
    193 static int sysctl_hw_usermem(SYSCTLFN_PROTO);
    194 static int sysctl_hw_cnmagic(SYSCTLFN_PROTO);
    195 
    196 static u_int sysctl_map_flags(const u_int *, u_int);
    197 static void fill_kproc2(struct proc *, struct kinfo_proc2 *);
    198 static void fill_lwp(struct lwp *l, struct kinfo_lwp *kl);
    199 static void fill_file(struct kinfo_file *, const struct file *, struct proc *,
    200 		      int);
    201 
    202 /*
    203  * ********************************************************************
    204  * section 1: setup routines
    205  * ********************************************************************
    206  * These functions are stuffed into a link set for sysctl setup
    207  * functions. They're never called or referenced from anywhere else.
    208  * ********************************************************************
    209  */
    210 
    211 /*
    212  * sets up the base nodes...
    213  */
    214 SYSCTL_SETUP(sysctl_root_setup, "sysctl base setup")
    215 {
    216 
    217 	sysctl_createv(clog, 0, NULL, NULL,
    218 		       CTLFLAG_PERMANENT,
    219 		       CTLTYPE_NODE, "kern",
    220 		       SYSCTL_DESCR("High kernel"),
    221 		       NULL, 0, NULL, 0,
    222 		       CTL_KERN, CTL_EOL);
    223 	sysctl_createv(clog, 0, NULL, NULL,
    224 		       CTLFLAG_PERMANENT,
    225 		       CTLTYPE_NODE, "vm",
    226 		       SYSCTL_DESCR("Virtual memory"),
    227 		       NULL, 0, NULL, 0,
    228 		       CTL_VM, CTL_EOL);
    229 	sysctl_createv(clog, 0, NULL, NULL,
    230 		       CTLFLAG_PERMANENT,
    231 		       CTLTYPE_NODE, "vfs",
    232 		       SYSCTL_DESCR("Filesystem"),
    233 		       NULL, 0, NULL, 0,
    234 		       CTL_VFS, CTL_EOL);
    235 	sysctl_createv(clog, 0, NULL, NULL,
    236 		       CTLFLAG_PERMANENT,
    237 		       CTLTYPE_NODE, "net",
    238 		       SYSCTL_DESCR("Networking"),
    239 		       NULL, 0, NULL, 0,
    240 		       CTL_NET, CTL_EOL);
    241 	sysctl_createv(clog, 0, NULL, NULL,
    242 		       CTLFLAG_PERMANENT,
    243 		       CTLTYPE_NODE, "debug",
    244 		       SYSCTL_DESCR("Debugging"),
    245 		       NULL, 0, NULL, 0,
    246 		       CTL_DEBUG, CTL_EOL);
    247 	sysctl_createv(clog, 0, NULL, NULL,
    248 		       CTLFLAG_PERMANENT,
    249 		       CTLTYPE_NODE, "hw",
    250 		       SYSCTL_DESCR("Generic CPU, I/O"),
    251 		       NULL, 0, NULL, 0,
    252 		       CTL_HW, CTL_EOL);
    253 	sysctl_createv(clog, 0, NULL, NULL,
    254 		       CTLFLAG_PERMANENT,
    255 		       CTLTYPE_NODE, "machdep",
    256 		       SYSCTL_DESCR("Machine dependent"),
    257 		       NULL, 0, NULL, 0,
    258 		       CTL_MACHDEP, CTL_EOL);
    259 	/*
    260 	 * this node is inserted so that the sysctl nodes in libc can
    261 	 * operate.
    262 	 */
    263 	sysctl_createv(clog, 0, NULL, NULL,
    264 		       CTLFLAG_PERMANENT,
    265 		       CTLTYPE_NODE, "user",
    266 		       SYSCTL_DESCR("User-level"),
    267 		       NULL, 0, NULL, 0,
    268 		       CTL_USER, CTL_EOL);
    269 	sysctl_createv(clog, 0, NULL, NULL,
    270 		       CTLFLAG_PERMANENT,
    271 		       CTLTYPE_NODE, "ddb",
    272 		       SYSCTL_DESCR("In-kernel debugger"),
    273 		       NULL, 0, NULL, 0,
    274 		       CTL_DDB, CTL_EOL);
    275 	sysctl_createv(clog, 0, NULL, NULL,
    276 		       CTLFLAG_PERMANENT,
    277 		       CTLTYPE_NODE, "proc",
    278 		       SYSCTL_DESCR("Per-process"),
    279 		       NULL, 0, NULL, 0,
    280 		       CTL_PROC, CTL_EOL);
    281 	sysctl_createv(clog, 0, NULL, NULL,
    282 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
    283 		       CTLTYPE_NODE, "vendor",
    284 		       SYSCTL_DESCR("Vendor specific"),
    285 		       NULL, 0, NULL, 0,
    286 		       CTL_VENDOR, CTL_EOL);
    287 	sysctl_createv(clog, 0, NULL, NULL,
    288 		       CTLFLAG_PERMANENT,
    289 		       CTLTYPE_NODE, "emul",
    290 		       SYSCTL_DESCR("Emulation settings"),
    291 		       NULL, 0, NULL, 0,
    292 		       CTL_EMUL, CTL_EOL);
    293 	sysctl_createv(clog, 0, NULL, NULL,
    294 		       CTLFLAG_PERMANENT,
    295 		       CTLTYPE_NODE, "security",
    296 		       SYSCTL_DESCR("Security"),
    297 		       NULL, 0, NULL, 0,
    298 		       CTL_SECURITY, CTL_EOL);
    299 }
    300 
    301 /*
    302  * this setup routine is a replacement for kern_sysctl()
    303  */
    304 SYSCTL_SETUP(sysctl_kern_setup, "sysctl kern subtree setup")
    305 {
    306 	extern int kern_logsigexit;	/* defined in kern/kern_sig.c */
    307 	extern fixpt_t ccpu;		/* defined in kern/kern_synch.c */
    308 	extern int dumponpanic;		/* defined in kern/subr_prf.c */
    309 	const struct sysctlnode *rnode;
    310 
    311 	sysctl_createv(clog, 0, NULL, NULL,
    312 		       CTLFLAG_PERMANENT,
    313 		       CTLTYPE_NODE, "kern", NULL,
    314 		       NULL, 0, NULL, 0,
    315 		       CTL_KERN, CTL_EOL);
    316 
    317 	sysctl_createv(clog, 0, NULL, NULL,
    318 		       CTLFLAG_PERMANENT,
    319 		       CTLTYPE_STRING, "ostype",
    320 		       SYSCTL_DESCR("Operating system type"),
    321 		       NULL, 0, &ostype, 0,
    322 		       CTL_KERN, KERN_OSTYPE, CTL_EOL);
    323 	sysctl_createv(clog, 0, NULL, NULL,
    324 		       CTLFLAG_PERMANENT,
    325 		       CTLTYPE_STRING, "osrelease",
    326 		       SYSCTL_DESCR("Operating system release"),
    327 		       NULL, 0, &osrelease, 0,
    328 		       CTL_KERN, KERN_OSRELEASE, CTL_EOL);
    329 	sysctl_createv(clog, 0, NULL, NULL,
    330 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    331 		       CTLTYPE_INT, "osrevision",
    332 		       SYSCTL_DESCR("Operating system revision"),
    333 		       NULL, __NetBSD_Version__, NULL, 0,
    334 		       CTL_KERN, KERN_OSREV, CTL_EOL);
    335 	sysctl_createv(clog, 0, NULL, NULL,
    336 		       CTLFLAG_PERMANENT,
    337 		       CTLTYPE_STRING, "version",
    338 		       SYSCTL_DESCR("Kernel version"),
    339 		       NULL, 0, &version, 0,
    340 		       CTL_KERN, KERN_VERSION, CTL_EOL);
    341 	sysctl_createv(clog, 0, NULL, NULL,
    342 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
    343 		       CTLTYPE_INT, "maxvnodes",
    344 		       SYSCTL_DESCR("Maximum number of vnodes"),
    345 		       sysctl_kern_maxvnodes, 0, NULL, 0,
    346 		       CTL_KERN, KERN_MAXVNODES, CTL_EOL);
    347 	sysctl_createv(clog, 0, NULL, NULL,
    348 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
    349 		       CTLTYPE_INT, "maxproc",
    350 		       SYSCTL_DESCR("Maximum number of simultaneous processes"),
    351 		       sysctl_kern_maxproc, 0, NULL, 0,
    352 		       CTL_KERN, KERN_MAXPROC, CTL_EOL);
    353 	sysctl_createv(clog, 0, NULL, NULL,
    354 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
    355 		       CTLTYPE_INT, "maxfiles",
    356 		       SYSCTL_DESCR("Maximum number of open files"),
    357 		       NULL, 0, &maxfiles, 0,
    358 		       CTL_KERN, KERN_MAXFILES, CTL_EOL);
    359 	sysctl_createv(clog, 0, NULL, NULL,
    360 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    361 		       CTLTYPE_INT, "argmax",
    362 		       SYSCTL_DESCR("Maximum number of bytes of arguments to "
    363 				    "execve(2)"),
    364 		       NULL, ARG_MAX, NULL, 0,
    365 		       CTL_KERN, KERN_ARGMAX, CTL_EOL);
    366 	sysctl_createv(clog, 0, NULL, NULL,
    367 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
    368 		       CTLTYPE_STRING, "hostname",
    369 		       SYSCTL_DESCR("System hostname"),
    370 		       sysctl_setlen, 0, &hostname, MAXHOSTNAMELEN,
    371 		       CTL_KERN, KERN_HOSTNAME, CTL_EOL);
    372 	sysctl_createv(clog, 0, NULL, NULL,
    373 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE|CTLFLAG_HEX,
    374 		       CTLTYPE_INT, "hostid",
    375 		       SYSCTL_DESCR("System host ID number"),
    376 		       sysctl_kern_hostid, 0, NULL, 0,
    377 		       CTL_KERN, KERN_HOSTID, CTL_EOL);
    378 	sysctl_createv(clog, 0, NULL, NULL,
    379 		       CTLFLAG_PERMANENT,
    380 		       CTLTYPE_STRUCT, "clockrate",
    381 		       SYSCTL_DESCR("Kernel clock rates"),
    382 		       sysctl_kern_clockrate, 0, NULL,
    383 		       sizeof(struct clockinfo),
    384 		       CTL_KERN, KERN_CLOCKRATE, CTL_EOL);
    385 	sysctl_createv(clog, 0, NULL, NULL,
    386 		       CTLFLAG_PERMANENT,
    387 		       CTLTYPE_INT, "hardclock_ticks",
    388 		       SYSCTL_DESCR("Number of hardclock ticks"),
    389 		       NULL, 0, &hardclock_ticks, sizeof(hardclock_ticks),
    390 		       CTL_KERN, KERN_HARDCLOCK_TICKS, CTL_EOL);
    391 	sysctl_createv(clog, 0, NULL, NULL,
    392 		       CTLFLAG_PERMANENT,
    393 		       CTLTYPE_STRUCT, "vnode",
    394 		       SYSCTL_DESCR("System vnode table"),
    395 		       sysctl_kern_vnode, 0, NULL, 0,
    396 		       CTL_KERN, KERN_VNODE, CTL_EOL);
    397 	sysctl_createv(clog, 0, NULL, NULL,
    398 		       CTLFLAG_PERMANENT,
    399 		       CTLTYPE_STRUCT, "file",
    400 		       SYSCTL_DESCR("System open file table"),
    401 		       sysctl_kern_file, 0, NULL, 0,
    402 		       CTL_KERN, KERN_FILE, CTL_EOL);
    403 #ifndef GPROF
    404 	sysctl_createv(clog, 0, NULL, NULL,
    405 		       CTLFLAG_PERMANENT,
    406 		       CTLTYPE_NODE, "profiling",
    407 		       SYSCTL_DESCR("Profiling information (not available)"),
    408 		       sysctl_notavail, 0, NULL, 0,
    409 		       CTL_KERN, KERN_PROF, CTL_EOL);
    410 #endif
    411 	sysctl_createv(clog, 0, NULL, NULL,
    412 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    413 		       CTLTYPE_INT, "posix1version",
    414 		       SYSCTL_DESCR("Version of ISO/IEC 9945 (POSIX 1003.1) "
    415 				    "with which the operating system attempts "
    416 				    "to comply"),
    417 		       NULL, _POSIX_VERSION, NULL, 0,
    418 		       CTL_KERN, KERN_POSIX1, CTL_EOL);
    419 	sysctl_createv(clog, 0, NULL, NULL,
    420 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    421 		       CTLTYPE_INT, "ngroups",
    422 		       SYSCTL_DESCR("Maximum number of supplemental groups"),
    423 		       NULL, NGROUPS_MAX, NULL, 0,
    424 		       CTL_KERN, KERN_NGROUPS, CTL_EOL);
    425 	sysctl_createv(clog, 0, NULL, NULL,
    426 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    427 		       CTLTYPE_INT, "job_control",
    428 		       SYSCTL_DESCR("Whether job control is available"),
    429 		       NULL, 1, NULL, 0,
    430 		       CTL_KERN, KERN_JOB_CONTROL, CTL_EOL);
    431 	sysctl_createv(clog, 0, NULL, NULL,
    432 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    433 		       CTLTYPE_INT, "saved_ids",
    434 		       SYSCTL_DESCR("Whether POSIX saved set-group/user ID is "
    435 				    "available"), NULL,
    436 #ifdef _POSIX_SAVED_IDS
    437 		       1,
    438 #else /* _POSIX_SAVED_IDS */
    439 		       0,
    440 #endif /* _POSIX_SAVED_IDS */
    441 		       NULL, 0, CTL_KERN, KERN_SAVED_IDS, CTL_EOL);
    442 	sysctl_createv(clog, 0, NULL, NULL,
    443 		       CTLFLAG_PERMANENT,
    444 		       CTLTYPE_STRUCT, "boottime",
    445 		       SYSCTL_DESCR("System boot time"),
    446 		       NULL, 0, &boottime, sizeof(boottime),
    447 		       CTL_KERN, KERN_BOOTTIME, CTL_EOL);
    448 	sysctl_createv(clog, 0, NULL, NULL,
    449 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
    450 		       CTLTYPE_STRING, "domainname",
    451 		       SYSCTL_DESCR("YP domain name"),
    452 		       sysctl_setlen, 0, &domainname, MAXHOSTNAMELEN,
    453 		       CTL_KERN, KERN_DOMAINNAME, CTL_EOL);
    454 	sysctl_createv(clog, 0, NULL, NULL,
    455 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    456 		       CTLTYPE_INT, "maxpartitions",
    457 		       SYSCTL_DESCR("Maximum number of partitions allowed per "
    458 				    "disk"),
    459 		       NULL, MAXPARTITIONS, NULL, 0,
    460 		       CTL_KERN, KERN_MAXPARTITIONS, CTL_EOL);
    461 	sysctl_createv(clog, 0, NULL, NULL,
    462 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    463 		       CTLTYPE_INT, "rawpartition",
    464 		       SYSCTL_DESCR("Raw partition of a disk"),
    465 		       NULL, RAW_PART, NULL, 0,
    466 		       CTL_KERN, KERN_RAWPARTITION, CTL_EOL);
    467 	sysctl_createv(clog, 0, NULL, NULL,
    468 		       CTLFLAG_PERMANENT,
    469 		       CTLTYPE_STRUCT, "timex", NULL,
    470 		       sysctl_notavail, 0, NULL, 0,
    471 		       CTL_KERN, KERN_TIMEX, CTL_EOL);
    472 	sysctl_createv(clog, 0, NULL, NULL,
    473 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
    474 		       CTLTYPE_INT, "rtc_offset",
    475 		       SYSCTL_DESCR("Offset of real time clock from UTC in "
    476 				    "minutes"),
    477 		       sysctl_kern_rtc_offset, 0, &rtc_offset, 0,
    478 		       CTL_KERN, KERN_RTC_OFFSET, CTL_EOL);
    479 	sysctl_createv(clog, 0, NULL, NULL,
    480 		       CTLFLAG_PERMANENT,
    481 		       CTLTYPE_STRING, "root_device",
    482 		       SYSCTL_DESCR("Name of the root device"),
    483 		       sysctl_root_device, 0, NULL, 0,
    484 		       CTL_KERN, KERN_ROOT_DEVICE, CTL_EOL);
    485 	sysctl_createv(clog, 0, NULL, NULL,
    486 		       CTLFLAG_PERMANENT,
    487 		       CTLTYPE_INT, "msgbufsize",
    488 		       SYSCTL_DESCR("Size of the kernel message buffer"),
    489 		       sysctl_msgbuf, 0, NULL, 0,
    490 		       CTL_KERN, KERN_MSGBUFSIZE, CTL_EOL);
    491 	sysctl_createv(clog, 0, NULL, NULL,
    492 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    493 		       CTLTYPE_INT, "fsync",
    494 		       SYSCTL_DESCR("Whether the POSIX 1003.1b File "
    495 				    "Synchronization Option is available on "
    496 				    "this system"),
    497 		       NULL, 1, NULL, 0,
    498 		       CTL_KERN, KERN_FSYNC, CTL_EOL);
    499 	sysctl_createv(clog, 0, NULL, NULL,
    500 		       CTLFLAG_PERMANENT,
    501 		       CTLTYPE_NODE, "ipc",
    502 		       SYSCTL_DESCR("SysV IPC options"),
    503 		       NULL, 0, NULL, 0,
    504 		       CTL_KERN, KERN_SYSVIPC, CTL_EOL);
    505 	sysctl_createv(clog, 0, NULL, NULL,
    506 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    507 		       CTLTYPE_INT, "sysvmsg",
    508 		       SYSCTL_DESCR("System V style message support available"),
    509 		       NULL,
    510 #ifdef SYSVMSG
    511 		       1,
    512 #else /* SYSVMSG */
    513 		       0,
    514 #endif /* SYSVMSG */
    515 		       NULL, 0, CTL_KERN, KERN_SYSVIPC, KERN_SYSVIPC_MSG, CTL_EOL);
    516 	sysctl_createv(clog, 0, NULL, NULL,
    517 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    518 		       CTLTYPE_INT, "sysvsem",
    519 		       SYSCTL_DESCR("System V style semaphore support "
    520 				    "available"), NULL,
    521 #ifdef SYSVSEM
    522 		       1,
    523 #else /* SYSVSEM */
    524 		       0,
    525 #endif /* SYSVSEM */
    526 		       NULL, 0, CTL_KERN, KERN_SYSVIPC, KERN_SYSVIPC_SEM, CTL_EOL);
    527 	sysctl_createv(clog, 0, NULL, NULL,
    528 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    529 		       CTLTYPE_INT, "sysvshm",
    530 		       SYSCTL_DESCR("System V style shared memory support "
    531 				    "available"), NULL,
    532 #ifdef SYSVSHM
    533 		       1,
    534 #else /* SYSVSHM */
    535 		       0,
    536 #endif /* SYSVSHM */
    537 		       NULL, 0, CTL_KERN, KERN_SYSVIPC, KERN_SYSVIPC_SHM, CTL_EOL);
    538 	sysctl_createv(clog, 0, NULL, NULL,
    539 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    540 		       CTLTYPE_INT, "synchronized_io",
    541 		       SYSCTL_DESCR("Whether the POSIX 1003.1b Synchronized "
    542 				    "I/O Option is available on this system"),
    543 		       NULL, 1, NULL, 0,
    544 		       CTL_KERN, KERN_SYNCHRONIZED_IO, CTL_EOL);
    545 	sysctl_createv(clog, 0, NULL, NULL,
    546 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    547 		       CTLTYPE_INT, "iov_max",
    548 		       SYSCTL_DESCR("Maximum number of iovec structures per "
    549 				    "process"),
    550 		       NULL, IOV_MAX, NULL, 0,
    551 		       CTL_KERN, KERN_IOV_MAX, CTL_EOL);
    552 	sysctl_createv(clog, 0, NULL, NULL,
    553 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    554 		       CTLTYPE_INT, "mapped_files",
    555 		       SYSCTL_DESCR("Whether the POSIX 1003.1b Memory Mapped "
    556 				    "Files Option is available on this system"),
    557 		       NULL, 1, NULL, 0,
    558 		       CTL_KERN, KERN_MAPPED_FILES, CTL_EOL);
    559 	sysctl_createv(clog, 0, NULL, NULL,
    560 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    561 		       CTLTYPE_INT, "memlock",
    562 		       SYSCTL_DESCR("Whether the POSIX 1003.1b Process Memory "
    563 				    "Locking Option is available on this "
    564 				    "system"),
    565 		       NULL, 1, NULL, 0,
    566 		       CTL_KERN, KERN_MEMLOCK, CTL_EOL);
    567 	sysctl_createv(clog, 0, NULL, NULL,
    568 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    569 		       CTLTYPE_INT, "memlock_range",
    570 		       SYSCTL_DESCR("Whether the POSIX 1003.1b Range Memory "
    571 				    "Locking Option is available on this "
    572 				    "system"),
    573 		       NULL, 1, NULL, 0,
    574 		       CTL_KERN, KERN_MEMLOCK_RANGE, CTL_EOL);
    575 	sysctl_createv(clog, 0, NULL, NULL,
    576 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    577 		       CTLTYPE_INT, "memory_protection",
    578 		       SYSCTL_DESCR("Whether the POSIX 1003.1b Memory "
    579 				    "Protection Option is available on this "
    580 				    "system"),
    581 		       NULL, 1, NULL, 0,
    582 		       CTL_KERN, KERN_MEMORY_PROTECTION, CTL_EOL);
    583 	sysctl_createv(clog, 0, NULL, NULL,
    584 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    585 		       CTLTYPE_INT, "login_name_max",
    586 		       SYSCTL_DESCR("Maximum login name length"),
    587 		       NULL, LOGIN_NAME_MAX, NULL, 0,
    588 		       CTL_KERN, KERN_LOGIN_NAME_MAX, CTL_EOL);
    589 	sysctl_createv(clog, 0, NULL, NULL,
    590 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
    591 		       CTLTYPE_STRING, "defcorename",
    592 		       SYSCTL_DESCR("Default core file name"),
    593 		       sysctl_kern_defcorename, 0, defcorename, MAXPATHLEN,
    594 		       CTL_KERN, KERN_DEFCORENAME, CTL_EOL);
    595 	sysctl_createv(clog, 0, NULL, NULL,
    596 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
    597 		       CTLTYPE_INT, "logsigexit",
    598 		       SYSCTL_DESCR("Log process exit when caused by signals"),
    599 		       NULL, 0, &kern_logsigexit, 0,
    600 		       CTL_KERN, KERN_LOGSIGEXIT, CTL_EOL);
    601 	sysctl_createv(clog, 0, NULL, NULL,
    602 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    603 		       CTLTYPE_INT, "fscale",
    604 		       SYSCTL_DESCR("Kernel fixed-point scale factor"),
    605 		       NULL, FSCALE, NULL, 0,
    606 		       CTL_KERN, KERN_FSCALE, CTL_EOL);
    607 	sysctl_createv(clog, 0, NULL, NULL,
    608 		       CTLFLAG_PERMANENT,
    609 		       CTLTYPE_INT, "ccpu",
    610 		       SYSCTL_DESCR("Scheduler exponential decay value"),
    611 		       NULL, 0, &ccpu, 0,
    612 		       CTL_KERN, KERN_CCPU, CTL_EOL);
    613 	sysctl_createv(clog, 0, NULL, NULL,
    614 		       CTLFLAG_PERMANENT,
    615 		       CTLTYPE_STRUCT, "cp_time",
    616 		       SYSCTL_DESCR("Clock ticks spent in different CPU states"),
    617 		       sysctl_kern_cptime, 0, NULL, 0,
    618 		       CTL_KERN, KERN_CP_TIME, CTL_EOL);
    619 	sysctl_createv(clog, 0, NULL, NULL,
    620 		       CTLFLAG_PERMANENT,
    621 		       CTLTYPE_INT, "msgbuf",
    622 		       SYSCTL_DESCR("Kernel message buffer"),
    623 		       sysctl_msgbuf, 0, NULL, 0,
    624 		       CTL_KERN, KERN_MSGBUF, CTL_EOL);
    625 	sysctl_createv(clog, 0, NULL, NULL,
    626 		       CTLFLAG_PERMANENT,
    627 		       CTLTYPE_STRUCT, "consdev",
    628 		       SYSCTL_DESCR("Console device"),
    629 		       sysctl_consdev, 0, NULL, sizeof(dev_t),
    630 		       CTL_KERN, KERN_CONSDEV, CTL_EOL);
    631 #if NPTY > 0
    632 	sysctl_createv(clog, 0, NULL, NULL,
    633 		       CTLFLAG_PERMANENT,
    634 		       CTLTYPE_INT, "maxptys",
    635 		       SYSCTL_DESCR("Maximum number of pseudo-ttys"),
    636 		       sysctl_kern_maxptys, 0, NULL, 0,
    637 		       CTL_KERN, KERN_MAXPTYS, CTL_EOL);
    638 #endif /* NPTY > 0 */
    639 	sysctl_createv(clog, 0, NULL, NULL,
    640 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    641 		       CTLTYPE_INT, "maxphys",
    642 		       SYSCTL_DESCR("Maximum raw I/O transfer size"),
    643 		       NULL, MAXPHYS, NULL, 0,
    644 		       CTL_KERN, KERN_MAXPHYS, CTL_EOL);
    645 	sysctl_createv(clog, 0, NULL, NULL,
    646 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
    647 		       CTLTYPE_INT, "sbmax",
    648 		       SYSCTL_DESCR("Maximum socket buffer size"),
    649 		       sysctl_kern_sbmax, 0, NULL, 0,
    650 		       CTL_KERN, KERN_SBMAX, CTL_EOL);
    651 	sysctl_createv(clog, 0, NULL, NULL,
    652 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    653 		       CTLTYPE_INT, "monotonic_clock",
    654 		       SYSCTL_DESCR("Implementation version of the POSIX "
    655 				    "1003.1b Monotonic Clock Option"),
    656 		       /* XXX _POSIX_VERSION */
    657 		       NULL, _POSIX_MONOTONIC_CLOCK, NULL, 0,
    658 		       CTL_KERN, KERN_MONOTONIC_CLOCK, CTL_EOL);
    659 	sysctl_createv(clog, 0, NULL, NULL,
    660 		       CTLFLAG_PERMANENT,
    661 		       CTLTYPE_INT, "urandom",
    662 		       SYSCTL_DESCR("Random integer value"),
    663 		       sysctl_kern_urnd, 0, NULL, 0,
    664 		       CTL_KERN, KERN_URND, CTL_EOL);
    665 	sysctl_createv(clog, 0, NULL, NULL,
    666 		       CTLFLAG_PERMANENT,
    667 		       CTLTYPE_INT, "arandom",
    668 		       SYSCTL_DESCR("n bytes of random data"),
    669 		       sysctl_kern_arnd, 0, NULL, 0,
    670 		       CTL_KERN, KERN_ARND, CTL_EOL);
    671 	sysctl_createv(clog, 0, NULL, NULL,
    672 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    673 		       CTLTYPE_INT, "labelsector",
    674 		       SYSCTL_DESCR("Sector number containing the disklabel"),
    675 		       NULL, LABELSECTOR, NULL, 0,
    676 		       CTL_KERN, KERN_LABELSECTOR, CTL_EOL);
    677 	sysctl_createv(clog, 0, NULL, NULL,
    678 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    679 		       CTLTYPE_INT, "labeloffset",
    680 		       SYSCTL_DESCR("Offset of the disklabel within the "
    681 				    "sector"),
    682 		       NULL, LABELOFFSET, NULL, 0,
    683 		       CTL_KERN, KERN_LABELOFFSET, CTL_EOL);
    684 	sysctl_createv(clog, 0, NULL, NULL,
    685 		       CTLFLAG_PERMANENT,
    686 		       CTLTYPE_NODE, "lwp",
    687 		       SYSCTL_DESCR("System-wide LWP information"),
    688 		       sysctl_kern_lwp, 0, NULL, 0,
    689 		       CTL_KERN, KERN_LWP, CTL_EOL);
    690 	sysctl_createv(clog, 0, NULL, NULL,
    691 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
    692 		       CTLTYPE_INT, "forkfsleep",
    693 		       SYSCTL_DESCR("Milliseconds to sleep on fork failure due "
    694 				    "to process limits"),
    695 		       sysctl_kern_forkfsleep, 0, NULL, 0,
    696 		       CTL_KERN, KERN_FORKFSLEEP, CTL_EOL);
    697 	sysctl_createv(clog, 0, NULL, NULL,
    698 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    699 		       CTLTYPE_INT, "posix_threads",
    700 		       SYSCTL_DESCR("Version of IEEE Std 1003.1 and its "
    701 				    "Threads option to which the system "
    702 				    "attempts to conform"),
    703 		       /* XXX _POSIX_VERSION */
    704 		       NULL, _POSIX_THREADS, NULL, 0,
    705 		       CTL_KERN, KERN_POSIX_THREADS, CTL_EOL);
    706 	sysctl_createv(clog, 0, NULL, NULL,
    707 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    708 		       CTLTYPE_INT, "posix_semaphores",
    709 		       SYSCTL_DESCR("Version of IEEE Std 1003.1 and its "
    710 				    "Semaphores option to which the system "
    711 				    "attempts to conform"), NULL,
    712 #ifdef P1003_1B_SEMAPHORE
    713 		       200112,
    714 #else /* P1003_1B_SEMAPHORE */
    715 		       0,
    716 #endif /* P1003_1B_SEMAPHORE */
    717 		       NULL, 0, CTL_KERN, KERN_POSIX_SEMAPHORES, CTL_EOL);
    718 	sysctl_createv(clog, 0, NULL, NULL,
    719 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    720 		       CTLTYPE_INT, "posix_barriers",
    721 		       SYSCTL_DESCR("Version of IEEE Std 1003.1 and its "
    722 				    "Barriers option to which the system "
    723 				    "attempts to conform"),
    724 		       /* XXX _POSIX_VERSION */
    725 		       NULL, _POSIX_BARRIERS, NULL, 0,
    726 		       CTL_KERN, KERN_POSIX_BARRIERS, CTL_EOL);
    727 	sysctl_createv(clog, 0, NULL, NULL,
    728 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    729 		       CTLTYPE_INT, "posix_timers",
    730 		       SYSCTL_DESCR("Version of IEEE Std 1003.1 and its "
    731 				    "Timers option to which the system "
    732 				    "attempts to conform"),
    733 		       /* XXX _POSIX_VERSION */
    734 		       NULL, _POSIX_TIMERS, NULL, 0,
    735 		       CTL_KERN, KERN_POSIX_TIMERS, CTL_EOL);
    736 	sysctl_createv(clog, 0, NULL, NULL,
    737 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    738 		       CTLTYPE_INT, "posix_spin_locks",
    739 		       SYSCTL_DESCR("Version of IEEE Std 1003.1 and its Spin "
    740 				    "Locks option to which the system attempts "
    741 				    "to conform"),
    742 		       /* XXX _POSIX_VERSION */
    743 		       NULL, _POSIX_SPIN_LOCKS, NULL, 0,
    744 		       CTL_KERN, KERN_POSIX_SPIN_LOCKS, CTL_EOL);
    745 	sysctl_createv(clog, 0, NULL, NULL,
    746 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    747 		       CTLTYPE_INT, "posix_reader_writer_locks",
    748 		       SYSCTL_DESCR("Version of IEEE Std 1003.1 and its "
    749 				    "Read-Write Locks option to which the "
    750 				    "system attempts to conform"),
    751 		       /* XXX _POSIX_VERSION */
    752 		       NULL, _POSIX_READER_WRITER_LOCKS, NULL, 0,
    753 		       CTL_KERN, KERN_POSIX_READER_WRITER_LOCKS, CTL_EOL);
    754 	sysctl_createv(clog, 0, NULL, NULL,
    755 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
    756 		       CTLTYPE_INT, "dump_on_panic",
    757 		       SYSCTL_DESCR("Perform a crash dump on system panic"),
    758 		       NULL, 0, &dumponpanic, 0,
    759 		       CTL_KERN, KERN_DUMP_ON_PANIC, CTL_EOL);
    760 #ifdef DIAGNOSTIC
    761 	sysctl_createv(clog, 0, NULL, NULL,
    762 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
    763 		       CTLTYPE_INT, "panic_now",
    764 		       SYSCTL_DESCR("Trigger a panic"),
    765 		       sysctl_kern_trigger_panic, 0, NULL, 0,
    766 		       CTL_KERN, CTL_CREATE, CTL_EOL);
    767 #endif
    768 	sysctl_createv(clog, 0, NULL, NULL,
    769 		       CTLFLAG_PERMANENT,
    770 		       CTLTYPE_INT, "root_partition",
    771 		       SYSCTL_DESCR("Root partition on the root device"),
    772 		       sysctl_kern_root_partition, 0, NULL, 0,
    773 		       CTL_KERN, KERN_ROOT_PARTITION, CTL_EOL);
    774 	sysctl_createv(clog, 0, NULL, NULL,
    775 		       CTLFLAG_PERMANENT,
    776 		       CTLTYPE_STRUCT, "drivers",
    777 		       SYSCTL_DESCR("List of all drivers with block and "
    778 				    "character device numbers"),
    779 		       sysctl_kern_drivers, 0, NULL, 0,
    780 		       CTL_KERN, KERN_DRIVERS, CTL_EOL);
    781 	sysctl_createv(clog, 0, NULL, NULL,
    782 		       CTLFLAG_PERMANENT,
    783 		       CTLTYPE_STRUCT, "file2",
    784 		       SYSCTL_DESCR("System open file table"),
    785 		       sysctl_kern_file2, 0, NULL, 0,
    786 		       CTL_KERN, KERN_FILE2, CTL_EOL);
    787 	sysctl_createv(clog, 0, NULL, NULL,
    788 		       CTLFLAG_PERMANENT,
    789 		       CTLTYPE_STRUCT, "cp_id",
    790 		       SYSCTL_DESCR("Mapping of CPU number to CPU id"),
    791 		       sysctl_kern_cpid, 0, NULL, 0,
    792 		       CTL_KERN, KERN_CP_ID, CTL_EOL);
    793 	sysctl_createv(clog, 0, NULL, &rnode,
    794 		       CTLFLAG_PERMANENT,
    795 		       CTLTYPE_NODE, "coredump",
    796 		       SYSCTL_DESCR("Coredump settings."),
    797 		       NULL, 0, NULL, 0,
    798 		       CTL_KERN, CTL_CREATE, CTL_EOL);
    799 	sysctl_createv(clog, 0, &rnode, &rnode,
    800 		       CTLFLAG_PERMANENT,
    801 		       CTLTYPE_NODE, "setid",
    802 		       SYSCTL_DESCR("Set-id processes' coredump settings."),
    803 		       NULL, 0, NULL, 0,
    804 		       CTL_CREATE, CTL_EOL);
    805 	sysctl_createv(clog, 0, &rnode, NULL,
    806 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
    807 		       CTLTYPE_INT, "dump",
    808 		       SYSCTL_DESCR("Allow set-id processes to dump core."),
    809 		       sysctl_security_setidcore, 0, &security_setidcore_dump,
    810 		       sizeof(security_setidcore_dump),
    811 		       CTL_CREATE, CTL_EOL);
    812 	sysctl_createv(clog, 0, &rnode, NULL,
    813 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
    814 		       CTLTYPE_STRING, "path",
    815 		       SYSCTL_DESCR("Path pattern for set-id coredumps."),
    816 		       sysctl_security_setidcorename, 0,
    817 		       &security_setidcore_path,
    818 		       sizeof(security_setidcore_path),
    819 		       CTL_CREATE, CTL_EOL);
    820 	sysctl_createv(clog, 0, &rnode, NULL,
    821 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
    822 		       CTLTYPE_INT, "owner",
    823 		       SYSCTL_DESCR("Owner id for set-id processes' cores."),
    824 		       sysctl_security_setidcore, 0, &security_setidcore_owner,
    825 		       0,
    826 		       CTL_CREATE, CTL_EOL);
    827 	sysctl_createv(clog, 0, &rnode, NULL,
    828 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
    829 		       CTLTYPE_INT, "group",
    830 		       SYSCTL_DESCR("Group id for set-id processes' cores."),
    831 		       sysctl_security_setidcore, 0, &security_setidcore_group,
    832 		       0,
    833 		       CTL_CREATE, CTL_EOL);
    834 	sysctl_createv(clog, 0, &rnode, NULL,
    835 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
    836 		       CTLTYPE_INT, "mode",
    837 		       SYSCTL_DESCR("Mode for set-id processes' cores."),
    838 		       sysctl_security_setidcore, 0, &security_setidcore_mode,
    839 		       0,
    840 		       CTL_CREATE, CTL_EOL);
    841 	sysctl_createv(clog, 0, NULL, NULL,
    842 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    843 		       CTLTYPE_INT, "no_sa_support",
    844 		       SYSCTL_DESCR("0 if the kernel supports SA, otherwise it doesn't"),
    845 		       NULL, 1, NULL, 0,
    846 		       CTL_KERN, CTL_CREATE, CTL_EOL);
    847 }
    848 
    849 SYSCTL_SETUP(sysctl_kern_proc_setup,
    850 	     "sysctl kern.proc/proc2/proc_args subtree setup")
    851 {
    852 
    853 	sysctl_createv(clog, 0, NULL, NULL,
    854 		       CTLFLAG_PERMANENT,
    855 		       CTLTYPE_NODE, "kern", NULL,
    856 		       NULL, 0, NULL, 0,
    857 		       CTL_KERN, CTL_EOL);
    858 
    859 	sysctl_createv(clog, 0, NULL, NULL,
    860 		       CTLFLAG_PERMANENT,
    861 		       CTLTYPE_NODE, "proc",
    862 		       SYSCTL_DESCR("System-wide process information"),
    863 		       sysctl_doeproc, 0, NULL, 0,
    864 		       CTL_KERN, KERN_PROC, CTL_EOL);
    865 	sysctl_createv(clog, 0, NULL, NULL,
    866 		       CTLFLAG_PERMANENT,
    867 		       CTLTYPE_NODE, "proc2",
    868 		       SYSCTL_DESCR("Machine-independent process information"),
    869 		       sysctl_doeproc, 0, NULL, 0,
    870 		       CTL_KERN, KERN_PROC2, CTL_EOL);
    871 	sysctl_createv(clog, 0, NULL, NULL,
    872 		       CTLFLAG_PERMANENT,
    873 		       CTLTYPE_NODE, "proc_args",
    874 		       SYSCTL_DESCR("Process argument information"),
    875 		       sysctl_kern_proc_args, 0, NULL, 0,
    876 		       CTL_KERN, KERN_PROC_ARGS, CTL_EOL);
    877 
    878 	/*
    879 	  "nodes" under these:
    880 
    881 	  KERN_PROC_ALL
    882 	  KERN_PROC_PID pid
    883 	  KERN_PROC_PGRP pgrp
    884 	  KERN_PROC_SESSION sess
    885 	  KERN_PROC_TTY tty
    886 	  KERN_PROC_UID uid
    887 	  KERN_PROC_RUID uid
    888 	  KERN_PROC_GID gid
    889 	  KERN_PROC_RGID gid
    890 
    891 	  all in all, probably not worth the effort...
    892 	*/
    893 }
    894 
    895 SYSCTL_SETUP(sysctl_hw_setup, "sysctl hw subtree setup")
    896 {
    897 	u_int u;
    898 	u_quad_t q;
    899 
    900 	sysctl_createv(clog, 0, NULL, NULL,
    901 		       CTLFLAG_PERMANENT,
    902 		       CTLTYPE_NODE, "hw", NULL,
    903 		       NULL, 0, NULL, 0,
    904 		       CTL_HW, CTL_EOL);
    905 
    906 	sysctl_createv(clog, 0, NULL, NULL,
    907 		       CTLFLAG_PERMANENT,
    908 		       CTLTYPE_STRING, "machine",
    909 		       SYSCTL_DESCR("Machine class"),
    910 		       NULL, 0, machine, 0,
    911 		       CTL_HW, HW_MACHINE, CTL_EOL);
    912 	sysctl_createv(clog, 0, NULL, NULL,
    913 		       CTLFLAG_PERMANENT,
    914 		       CTLTYPE_STRING, "model",
    915 		       SYSCTL_DESCR("Machine model"),
    916 		       NULL, 0, cpu_model, 0,
    917 		       CTL_HW, HW_MODEL, CTL_EOL);
    918 	sysctl_createv(clog, 0, NULL, NULL,
    919 		       CTLFLAG_PERMANENT,
    920 		       CTLTYPE_INT, "ncpu",
    921 		       SYSCTL_DESCR("Number of CPUs configured"),
    922 		       NULL, 0, &ncpu, 0,
    923 		       CTL_HW, HW_NCPU, CTL_EOL);
    924 	sysctl_createv(clog, 0, NULL, NULL,
    925 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    926 		       CTLTYPE_INT, "byteorder",
    927 		       SYSCTL_DESCR("System byte order"),
    928 		       NULL, BYTE_ORDER, NULL, 0,
    929 		       CTL_HW, HW_BYTEORDER, CTL_EOL);
    930 	u = ((u_int)physmem > (UINT_MAX / PAGE_SIZE)) ?
    931 		UINT_MAX : physmem * PAGE_SIZE;
    932 	sysctl_createv(clog, 0, NULL, NULL,
    933 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    934 		       CTLTYPE_INT, "physmem",
    935 		       SYSCTL_DESCR("Bytes of physical memory"),
    936 		       NULL, u, NULL, 0,
    937 		       CTL_HW, HW_PHYSMEM, CTL_EOL);
    938 	sysctl_createv(clog, 0, NULL, NULL,
    939 		       CTLFLAG_PERMANENT,
    940 		       CTLTYPE_INT, "usermem",
    941 		       SYSCTL_DESCR("Bytes of non-kernel memory"),
    942 		       sysctl_hw_usermem, 0, NULL, 0,
    943 		       CTL_HW, HW_USERMEM, CTL_EOL);
    944 	sysctl_createv(clog, 0, NULL, NULL,
    945 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    946 		       CTLTYPE_INT, "pagesize",
    947 		       SYSCTL_DESCR("Software page size"),
    948 		       NULL, PAGE_SIZE, NULL, 0,
    949 		       CTL_HW, HW_PAGESIZE, CTL_EOL);
    950 	sysctl_createv(clog, 0, NULL, NULL,
    951 		       CTLFLAG_PERMANENT,
    952 		       CTLTYPE_STRING, "machine_arch",
    953 		       SYSCTL_DESCR("Machine CPU class"),
    954 		       NULL, 0, machine_arch, 0,
    955 		       CTL_HW, HW_MACHINE_ARCH, CTL_EOL);
    956 	sysctl_createv(clog, 0, NULL, NULL,
    957 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    958 		       CTLTYPE_INT, "alignbytes",
    959 		       SYSCTL_DESCR("Alignment constraint for all possible "
    960 				    "data types"),
    961 		       NULL, ALIGNBYTES, NULL, 0,
    962 		       CTL_HW, HW_ALIGNBYTES, CTL_EOL);
    963 	sysctl_createv(clog, 0, NULL, NULL,
    964 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE|CTLFLAG_HEX,
    965 		       CTLTYPE_STRING, "cnmagic",
    966 		       SYSCTL_DESCR("Console magic key sequence"),
    967 		       sysctl_hw_cnmagic, 0, NULL, CNS_LEN,
    968 		       CTL_HW, HW_CNMAGIC, CTL_EOL);
    969 	q = (u_quad_t)physmem * PAGE_SIZE;
    970 	sysctl_createv(clog, 0, NULL, NULL,
    971 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    972 		       CTLTYPE_QUAD, "physmem64",
    973 		       SYSCTL_DESCR("Bytes of physical memory"),
    974 		       NULL, q, NULL, 0,
    975 		       CTL_HW, HW_PHYSMEM64, CTL_EOL);
    976 	sysctl_createv(clog, 0, NULL, NULL,
    977 		       CTLFLAG_PERMANENT,
    978 		       CTLTYPE_QUAD, "usermem64",
    979 		       SYSCTL_DESCR("Bytes of non-kernel memory"),
    980 		       sysctl_hw_usermem, 0, NULL, 0,
    981 		       CTL_HW, HW_USERMEM64, CTL_EOL);
    982 	sysctl_createv(clog, 0, NULL, NULL,
    983 		       CTLFLAG_PERMANENT,
    984 		       CTLTYPE_INT, "ncpuonline",
    985 		       SYSCTL_DESCR("Number of CPUs online"),
    986 		       NULL, 0, &ncpuonline, 0,
    987 		       CTL_HW, HW_NCPUONLINE, CTL_EOL);
    988 }
    989 
    990 #ifdef DEBUG
    991 /*
    992  * Debugging related system variables.
    993  */
    994 struct ctldebug /* debug0, */ /* debug1, */ debug2, debug3, debug4;
    995 struct ctldebug debug5, debug6, debug7, debug8, debug9;
    996 struct ctldebug debug10, debug11, debug12, debug13, debug14;
    997 struct ctldebug debug15, debug16, debug17, debug18, debug19;
    998 static struct ctldebug *debugvars[CTL_DEBUG_MAXID] = {
    999 	&debug0, &debug1, &debug2, &debug3, &debug4,
   1000 	&debug5, &debug6, &debug7, &debug8, &debug9,
   1001 	&debug10, &debug11, &debug12, &debug13, &debug14,
   1002 	&debug15, &debug16, &debug17, &debug18, &debug19,
   1003 };
   1004 
   1005 /*
   1006  * this setup routine is a replacement for debug_sysctl()
   1007  *
   1008  * note that it creates several nodes per defined debug variable
   1009  */
   1010 SYSCTL_SETUP(sysctl_debug_setup, "sysctl debug subtree setup")
   1011 {
   1012 	struct ctldebug *cdp;
   1013 	char nodename[20];
   1014 	int i;
   1015 
   1016 	/*
   1017 	 * two ways here:
   1018 	 *
   1019 	 * the "old" way (debug.name -> value) which was emulated by
   1020 	 * the sysctl(8) binary
   1021 	 *
   1022 	 * the new way, which the sysctl(8) binary was actually using
   1023 
   1024 	 node	debug
   1025 	 node	debug.0
   1026 	 string debug.0.name
   1027 	 int	debug.0.value
   1028 	 int	debug.name
   1029 
   1030 	 */
   1031 
   1032 	sysctl_createv(clog, 0, NULL, NULL,
   1033 		       CTLFLAG_PERMANENT,
   1034 		       CTLTYPE_NODE, "debug", NULL,
   1035 		       NULL, 0, NULL, 0,
   1036 		       CTL_DEBUG, CTL_EOL);
   1037 
   1038 	for (i = 0; i < CTL_DEBUG_MAXID; i++) {
   1039 		cdp = debugvars[i];
   1040 		if (cdp->debugname == NULL || cdp->debugvar == NULL)
   1041 			continue;
   1042 
   1043 		snprintf(nodename, sizeof(nodename), "debug%d", i);
   1044 		sysctl_createv(clog, 0, NULL, NULL,
   1045 			       CTLFLAG_PERMANENT|CTLFLAG_HIDDEN,
   1046 			       CTLTYPE_NODE, nodename, NULL,
   1047 			       NULL, 0, NULL, 0,
   1048 			       CTL_DEBUG, i, CTL_EOL);
   1049 		sysctl_createv(clog, 0, NULL, NULL,
   1050 			       CTLFLAG_PERMANENT|CTLFLAG_HIDDEN,
   1051 			       CTLTYPE_STRING, "name", NULL,
   1052 			       /*XXXUNCONST*/
   1053 			       NULL, 0, __UNCONST(cdp->debugname), 0,
   1054 			       CTL_DEBUG, i, CTL_DEBUG_NAME, CTL_EOL);
   1055 		sysctl_createv(clog, 0, NULL, NULL,
   1056 			       CTLFLAG_PERMANENT|CTLFLAG_HIDDEN,
   1057 			       CTLTYPE_INT, "value", NULL,
   1058 			       NULL, 0, cdp->debugvar, 0,
   1059 			       CTL_DEBUG, i, CTL_DEBUG_VALUE, CTL_EOL);
   1060 		sysctl_createv(clog, 0, NULL, NULL,
   1061 			       CTLFLAG_PERMANENT,
   1062 			       CTLTYPE_INT, cdp->debugname, NULL,
   1063 			       NULL, 0, cdp->debugvar, 0,
   1064 			       CTL_DEBUG, CTL_CREATE, CTL_EOL);
   1065 	}
   1066 }
   1067 #endif /* DEBUG */
   1068 
   1069 /*
   1070  * ********************************************************************
   1071  * section 2: private node-specific helper routines.
   1072  * ********************************************************************
   1073  */
   1074 
   1075 #ifdef DIAGNOSTIC
   1076 static int
   1077 sysctl_kern_trigger_panic(SYSCTLFN_ARGS)
   1078 {
   1079 	int newtrig, error;
   1080 	struct sysctlnode node;
   1081 
   1082 	newtrig = 0;
   1083 	node = *rnode;
   1084 	node.sysctl_data = &newtrig;
   1085 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
   1086 	if (error || newp == NULL)
   1087 		return (error);
   1088 
   1089 	if (newtrig != 0)
   1090 		panic("Panic triggered");
   1091 
   1092 	return (error);
   1093 }
   1094 #endif
   1095 
   1096 /*
   1097  * sysctl helper routine for kern.maxvnodes.  Drain vnodes if
   1098  * new value is lower than desiredvnodes and then calls reinit
   1099  * routines that needs to adjust to the new value.
   1100  */
   1101 static int
   1102 sysctl_kern_maxvnodes(SYSCTLFN_ARGS)
   1103 {
   1104 	int error, new_vnodes, old_vnodes;
   1105 	struct sysctlnode node;
   1106 
   1107 	new_vnodes = desiredvnodes;
   1108 	node = *rnode;
   1109 	node.sysctl_data = &new_vnodes;
   1110 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
   1111 	if (error || newp == NULL)
   1112 		return (error);
   1113 
   1114 	old_vnodes = desiredvnodes;
   1115 	desiredvnodes = new_vnodes;
   1116 	if (new_vnodes < old_vnodes) {
   1117 		error = vfs_drainvnodes(new_vnodes, l);
   1118 		if (error) {
   1119 			desiredvnodes = old_vnodes;
   1120 			return (error);
   1121 		}
   1122 	}
   1123 	vfs_reinit();
   1124 	nchreinit();
   1125 
   1126 	return (0);
   1127 }
   1128 
   1129 /*
   1130  * sysctl helper routine for rtc_offset - set time after changes
   1131  */
   1132 static int
   1133 sysctl_kern_rtc_offset(SYSCTLFN_ARGS)
   1134 {
   1135 	struct timespec ts, delta;
   1136 	int error, new_rtc_offset;
   1137 	struct sysctlnode node;
   1138 
   1139 	new_rtc_offset = rtc_offset;
   1140 	node = *rnode;
   1141 	node.sysctl_data = &new_rtc_offset;
   1142 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
   1143 	if (error || newp == NULL)
   1144 		return (error);
   1145 
   1146 	if (kauth_authorize_system(l->l_cred, KAUTH_SYSTEM_TIME,
   1147 	    KAUTH_REQ_SYSTEM_TIME_RTCOFFSET,
   1148 	    KAUTH_ARG(new_rtc_offset), NULL, NULL))
   1149 		return (EPERM);
   1150 	if (rtc_offset == new_rtc_offset)
   1151 		return (0);
   1152 
   1153 	/* if we change the offset, adjust the time */
   1154 	nanotime(&ts);
   1155 	delta.tv_sec = 60 * (new_rtc_offset - rtc_offset);
   1156 	delta.tv_nsec = 0;
   1157 	timespecadd(&ts, &delta, &ts);
   1158 	rtc_offset = new_rtc_offset;
   1159 	return (settime(l->l_proc, &ts));
   1160 }
   1161 
   1162 /*
   1163  * sysctl helper routine for kern.maxproc. Ensures that the new
   1164  * values are not too low or too high.
   1165  */
   1166 static int
   1167 sysctl_kern_maxproc(SYSCTLFN_ARGS)
   1168 {
   1169 	int error, nmaxproc;
   1170 	struct sysctlnode node;
   1171 
   1172 	nmaxproc = maxproc;
   1173 	node = *rnode;
   1174 	node.sysctl_data = &nmaxproc;
   1175 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
   1176 	if (error || newp == NULL)
   1177 		return (error);
   1178 
   1179 	if (nmaxproc < 0 || nmaxproc >= PID_MAX)
   1180 		return (EINVAL);
   1181 #ifdef __HAVE_CPU_MAXPROC
   1182 	if (nmaxproc > cpu_maxproc())
   1183 		return (EINVAL);
   1184 #endif
   1185 	maxproc = nmaxproc;
   1186 
   1187 	return (0);
   1188 }
   1189 
   1190 /*
   1191  * sysctl helper function for kern.hostid. The hostid is a long, but
   1192  * we export it as an int, so we need to give it a little help.
   1193  */
   1194 static int
   1195 sysctl_kern_hostid(SYSCTLFN_ARGS)
   1196 {
   1197 	int error, inthostid;
   1198 	struct sysctlnode node;
   1199 
   1200 	inthostid = hostid;  /* XXX assumes sizeof int <= sizeof long */
   1201 	node = *rnode;
   1202 	node.sysctl_data = &inthostid;
   1203 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
   1204 	if (error || newp == NULL)
   1205 		return (error);
   1206 
   1207 	hostid = (unsigned)inthostid;
   1208 
   1209 	return (0);
   1210 }
   1211 
   1212 /*
   1213  * sysctl helper function for kern.hostname and kern.domainnname.
   1214  * resets the relevant recorded length when the underlying name is
   1215  * changed.
   1216  */
   1217 static int
   1218 sysctl_setlen(SYSCTLFN_ARGS)
   1219 {
   1220 	int error;
   1221 
   1222 	error = sysctl_lookup(SYSCTLFN_CALL(rnode));
   1223 	if (error || newp == NULL)
   1224 		return (error);
   1225 
   1226 	switch (rnode->sysctl_num) {
   1227 	case KERN_HOSTNAME:
   1228 		hostnamelen = strlen((const char*)rnode->sysctl_data);
   1229 		break;
   1230 	case KERN_DOMAINNAME:
   1231 		domainnamelen = strlen((const char*)rnode->sysctl_data);
   1232 		break;
   1233 	}
   1234 
   1235 	return (0);
   1236 }
   1237 
   1238 /*
   1239  * sysctl helper routine for kern.clockrate. Assembles a struct on
   1240  * the fly to be returned to the caller.
   1241  */
   1242 static int
   1243 sysctl_kern_clockrate(SYSCTLFN_ARGS)
   1244 {
   1245 	struct clockinfo clkinfo;
   1246 	struct sysctlnode node;
   1247 
   1248 	clkinfo.tick = tick;
   1249 	clkinfo.tickadj = tickadj;
   1250 	clkinfo.hz = hz;
   1251 	clkinfo.profhz = profhz;
   1252 	clkinfo.stathz = stathz ? stathz : hz;
   1253 
   1254 	node = *rnode;
   1255 	node.sysctl_data = &clkinfo;
   1256 	return (sysctl_lookup(SYSCTLFN_CALL(&node)));
   1257 }
   1258 
   1259 
   1260 /*
   1261  * sysctl helper routine for kern.file pseudo-subtree.
   1262  */
   1263 static int
   1264 sysctl_kern_file(SYSCTLFN_ARGS)
   1265 {
   1266 	int error;
   1267 	size_t buflen;
   1268 	struct file *fp, *dp, *np, fbuf;
   1269 	char *start, *where;
   1270 
   1271 	start = where = oldp;
   1272 	buflen = *oldlenp;
   1273 	dp = NULL;
   1274 
   1275 	if (where == NULL) {
   1276 		/*
   1277 		 * overestimate by 10 files
   1278 		 */
   1279 		*oldlenp = sizeof(filehead) + (nfiles + 10) *
   1280 		    sizeof(struct file);
   1281 		return (0);
   1282 	}
   1283 
   1284 	/*
   1285 	 * first dcopyout filehead
   1286 	 */
   1287 	if (buflen < sizeof(filehead)) {
   1288 		*oldlenp = 0;
   1289 		return (0);
   1290 	}
   1291 	sysctl_unlock();
   1292 	error = dcopyout(l, &filehead, where, sizeof(filehead));
   1293 	if (error) {
   1294 	 	sysctl_relock();
   1295 		return error;
   1296 	}
   1297 	buflen -= sizeof(filehead);
   1298 	where += sizeof(filehead);
   1299 
   1300 	/*
   1301 	 * allocate dummy file descriptor to make position in list
   1302 	 */
   1303 	if ((dp = fgetdummy()) == NULL) {
   1304 	 	sysctl_relock();
   1305 		return ENOMEM;
   1306 	}
   1307 
   1308 	/*
   1309 	 * followed by an array of file structures
   1310 	 */
   1311 	mutex_enter(&filelist_lock);
   1312 	for (fp = LIST_FIRST(&filehead); fp != NULL; fp = np) {
   1313 	    	np = LIST_NEXT(fp, f_list);
   1314 	    	mutex_enter(&fp->f_lock);
   1315 	    	if (fp->f_count == 0) {
   1316 		    	mutex_exit(&fp->f_lock);
   1317 	    		continue;
   1318 		}
   1319 		/*
   1320 		 * XXX Need to prevent that from being an alternative way
   1321 		 * XXX to getting process information.
   1322 		 */
   1323 		if (kauth_authorize_generic(l->l_cred,
   1324 		    KAUTH_GENERIC_CANSEE, fp->f_cred) != 0) {
   1325 		    	mutex_exit(&fp->f_lock);
   1326 			continue;
   1327 		}
   1328 		if (buflen < sizeof(struct file)) {
   1329 			*oldlenp = where - start;
   1330 		    	mutex_exit(&fp->f_lock);
   1331 			error = ENOMEM;
   1332 			break;
   1333 		}
   1334 		memcpy(&fbuf, fp, sizeof(fbuf));
   1335 		LIST_INSERT_AFTER(fp, dp, f_list);
   1336 	    	mutex_exit(&fp->f_lock);
   1337 		mutex_exit(&filelist_lock);
   1338 		error = dcopyout(l, &fbuf, where, sizeof(fbuf));
   1339 		if (error) {
   1340 			mutex_enter(&filelist_lock);
   1341 			LIST_REMOVE(dp, f_list);
   1342 			break;
   1343 		}
   1344 		buflen -= sizeof(struct file);
   1345 		where += sizeof(struct file);
   1346 		mutex_enter(&filelist_lock);
   1347 		np = LIST_NEXT(dp, f_list);
   1348 		LIST_REMOVE(dp, f_list);
   1349 	}
   1350 	mutex_exit(&filelist_lock);
   1351 	*oldlenp = where - start;
   1352  	if (dp != NULL)
   1353 		fputdummy(dp);
   1354  	sysctl_relock();
   1355 	return (error);
   1356 }
   1357 
   1358 /*
   1359  * sysctl helper routine for kern.msgbufsize and kern.msgbuf. For the
   1360  * former it merely checks the message buffer is set up. For the latter,
   1361  * it also copies out the data if necessary.
   1362  */
   1363 static int
   1364 sysctl_msgbuf(SYSCTLFN_ARGS)
   1365 {
   1366 	char *where = oldp;
   1367 	size_t len, maxlen;
   1368 	long beg, end;
   1369 	extern kmutex_t log_lock;
   1370 	int error;
   1371 
   1372 	if (!msgbufenabled || msgbufp->msg_magic != MSG_MAGIC) {
   1373 		msgbufenabled = 0;
   1374 		return (ENXIO);
   1375 	}
   1376 
   1377 	switch (rnode->sysctl_num) {
   1378 	case KERN_MSGBUFSIZE: {
   1379 		struct sysctlnode node = *rnode;
   1380 		int msg_bufs = (int)msgbufp->msg_bufs;
   1381 		node.sysctl_data = &msg_bufs;
   1382 		return (sysctl_lookup(SYSCTLFN_CALL(&node)));
   1383 	}
   1384 	case KERN_MSGBUF:
   1385 		break;
   1386 	default:
   1387 		return (EOPNOTSUPP);
   1388 	}
   1389 
   1390 	if (newp != NULL)
   1391 		return (EPERM);
   1392 
   1393 	if (oldp == NULL) {
   1394 		/* always return full buffer size */
   1395 		*oldlenp = msgbufp->msg_bufs;
   1396 		return (0);
   1397 	}
   1398 
   1399 	sysctl_unlock();
   1400 
   1401 	/*
   1402 	 * First, copy from the write pointer to the end of
   1403 	 * message buffer.
   1404 	 */
   1405 	error = 0;
   1406 	mutex_spin_enter(&log_lock);
   1407 	maxlen = MIN(msgbufp->msg_bufs, *oldlenp);
   1408 	beg = msgbufp->msg_bufx;
   1409 	end = msgbufp->msg_bufs;
   1410 	mutex_spin_exit(&log_lock);
   1411 
   1412 	while (maxlen > 0) {
   1413 		len = MIN(end - beg, maxlen);
   1414 		if (len == 0)
   1415 			break;
   1416 		/* XXX unlocked, but hardly matters. */
   1417 		error = dcopyout(l, &msgbufp->msg_bufc[beg], where, len);
   1418 		if (error)
   1419 			break;
   1420 		where += len;
   1421 		maxlen -= len;
   1422 
   1423 		/*
   1424 		 * ... then, copy from the beginning of message buffer to
   1425 		 * the write pointer.
   1426 		 */
   1427 		beg = 0;
   1428 		end = msgbufp->msg_bufx;
   1429 	}
   1430 
   1431 	sysctl_relock();
   1432 	return (error);
   1433 }
   1434 
   1435 /*
   1436  * sysctl helper routine for kern.defcorename. In the case of a new
   1437  * string being assigned, check that it's not a zero-length string.
   1438  * (XXX the check in -current doesn't work, but do we really care?)
   1439  */
   1440 static int
   1441 sysctl_kern_defcorename(SYSCTLFN_ARGS)
   1442 {
   1443 	int error;
   1444 	char *newcorename;
   1445 	struct sysctlnode node;
   1446 
   1447 	newcorename = PNBUF_GET();
   1448 	node = *rnode;
   1449 	node.sysctl_data = &newcorename[0];
   1450 	memcpy(node.sysctl_data, rnode->sysctl_data, MAXPATHLEN);
   1451 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
   1452 	if (error || newp == NULL) {
   1453 		goto done;
   1454 	}
   1455 
   1456 	/*
   1457 	 * when sysctl_lookup() deals with a string, it's guaranteed
   1458 	 * to come back nul terminated. So there.  :)
   1459 	 */
   1460 	if (strlen(newcorename) == 0) {
   1461 		error = EINVAL;
   1462 	} else {
   1463 		memcpy(rnode->sysctl_data, node.sysctl_data, MAXPATHLEN);
   1464 		error = 0;
   1465 	}
   1466 done:
   1467 	PNBUF_PUT(newcorename);
   1468 	return error;
   1469 }
   1470 
   1471 /*
   1472  * sysctl helper routine for kern.cp_time node. Adds up cpu time
   1473  * across all cpus.
   1474  */
   1475 static int
   1476 sysctl_kern_cptime(SYSCTLFN_ARGS)
   1477 {
   1478 	struct sysctlnode node = *rnode;
   1479 	uint64_t *cp_time = NULL;
   1480 	int error, n = ncpu, i;
   1481 	struct cpu_info *ci;
   1482 	CPU_INFO_ITERATOR cii;
   1483 
   1484 	/*
   1485 	 * if you specifically pass a buffer that is the size of the
   1486 	 * sum, or if you are probing for the size, you get the "sum"
   1487 	 * of cp_time (and the size thereof) across all processors.
   1488 	 *
   1489 	 * alternately, you can pass an additional mib number and get
   1490 	 * cp_time for that particular processor.
   1491 	 */
   1492 	switch (namelen) {
   1493 	case 0:
   1494 		if (*oldlenp == sizeof(uint64_t) * CPUSTATES || oldp == NULL) {
   1495 			node.sysctl_size = sizeof(uint64_t) * CPUSTATES;
   1496 			n = -1; /* SUM */
   1497 		}
   1498 		else {
   1499 			node.sysctl_size = n * sizeof(uint64_t) * CPUSTATES;
   1500 			n = -2; /* ALL */
   1501 		}
   1502 		break;
   1503 	case 1:
   1504 		if (name[0] < 0 || name[0] >= n)
   1505 			return (ENOENT); /* ENOSUCHPROCESSOR */
   1506 		node.sysctl_size = sizeof(uint64_t) * CPUSTATES;
   1507 		n = name[0];
   1508 		/*
   1509 		 * adjust these so that sysctl_lookup() will be happy
   1510 		 */
   1511 		name++;
   1512 		namelen--;
   1513 		break;
   1514 	default:
   1515 		return (EINVAL);
   1516 	}
   1517 
   1518 	cp_time = kmem_alloc(node.sysctl_size, KM_SLEEP);
   1519 	if (cp_time == NULL)
   1520 		return (ENOMEM);
   1521 	node.sysctl_data = cp_time;
   1522 	memset(cp_time, 0, node.sysctl_size);
   1523 
   1524 	for (CPU_INFO_FOREACH(cii, ci)) {
   1525 		if (n <= 0) {
   1526 			for (i = 0; i < CPUSTATES; i++) {
   1527 				cp_time[i] += ci->ci_schedstate.spc_cp_time[i];
   1528 			}
   1529 		}
   1530 		/*
   1531 		 * if a specific processor was requested and we just
   1532 		 * did it, we're done here
   1533 		 */
   1534 		if (n == 0)
   1535 			break;
   1536 		/*
   1537 		 * if doing "all", skip to next cp_time set for next processor
   1538 		 */
   1539 		if (n == -2)
   1540 			cp_time += CPUSTATES;
   1541 		/*
   1542 		 * if we're doing a specific processor, we're one
   1543 		 * processor closer
   1544 		 */
   1545 		if (n > 0)
   1546 			n--;
   1547 	}
   1548 
   1549 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
   1550 	kmem_free(node.sysctl_data, node.sysctl_size);
   1551 	return (error);
   1552 }
   1553 
   1554 #if NPTY > 0
   1555 /*
   1556  * sysctl helper routine for kern.maxptys. Ensures that any new value
   1557  * is acceptable to the pty subsystem.
   1558  */
   1559 static int
   1560 sysctl_kern_maxptys(SYSCTLFN_ARGS)
   1561 {
   1562 	int pty_maxptys(int, int);		/* defined in kern/tty_pty.c */
   1563 	int error, xmax;
   1564 	struct sysctlnode node;
   1565 
   1566 	/* get current value of maxptys */
   1567 	xmax = pty_maxptys(0, 0);
   1568 
   1569 	node = *rnode;
   1570 	node.sysctl_data = &xmax;
   1571 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
   1572 	if (error || newp == NULL)
   1573 		return (error);
   1574 
   1575 	if (xmax != pty_maxptys(xmax, 1))
   1576 		return (EINVAL);
   1577 
   1578 	return (0);
   1579 }
   1580 #endif /* NPTY > 0 */
   1581 
   1582 /*
   1583  * sysctl helper routine for kern.sbmax. Basically just ensures that
   1584  * any new value is not too small.
   1585  */
   1586 static int
   1587 sysctl_kern_sbmax(SYSCTLFN_ARGS)
   1588 {
   1589 	int error, new_sbmax;
   1590 	struct sysctlnode node;
   1591 
   1592 	new_sbmax = sb_max;
   1593 	node = *rnode;
   1594 	node.sysctl_data = &new_sbmax;
   1595 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
   1596 	if (error || newp == NULL)
   1597 		return (error);
   1598 
   1599 	KERNEL_LOCK(1, NULL);
   1600 	error = sb_max_set(new_sbmax);
   1601 	KERNEL_UNLOCK_ONE(NULL);
   1602 
   1603 	return (error);
   1604 }
   1605 
   1606 /*
   1607  * sysctl helper routine for kern.urandom node. Picks a random number
   1608  * for you.
   1609  */
   1610 static int
   1611 sysctl_kern_urnd(SYSCTLFN_ARGS)
   1612 {
   1613 #if NRND > 0
   1614 	int v, rv;
   1615 
   1616 	KERNEL_LOCK(1, NULL);
   1617 	rv = rnd_extract_data(&v, sizeof(v), RND_EXTRACT_ANY);
   1618 	KERNEL_UNLOCK_ONE(NULL);
   1619 	if (rv == sizeof(v)) {
   1620 		struct sysctlnode node = *rnode;
   1621 		node.sysctl_data = &v;
   1622 		return (sysctl_lookup(SYSCTLFN_CALL(&node)));
   1623 	}
   1624 	else
   1625 		return (EIO);	/*XXX*/
   1626 #else
   1627 	return (EOPNOTSUPP);
   1628 #endif
   1629 }
   1630 
   1631 /*
   1632  * sysctl helper routine for kern.arandom node. Picks a random number
   1633  * for you.
   1634  */
   1635 static int
   1636 sysctl_kern_arnd(SYSCTLFN_ARGS)
   1637 {
   1638 #if NRND > 0
   1639 	int error;
   1640 	void *v;
   1641 	struct sysctlnode node = *rnode;
   1642 
   1643 	if (*oldlenp == 0)
   1644 		return 0;
   1645 	if (*oldlenp > 8192)
   1646 		return E2BIG;
   1647 
   1648 	v = kmem_alloc(*oldlenp, KM_SLEEP);
   1649 	arc4randbytes(v, *oldlenp);
   1650 	node.sysctl_data = v;
   1651 	node.sysctl_size = *oldlenp;
   1652 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
   1653 	kmem_free(v, *oldlenp);
   1654 	return error;
   1655 #else
   1656 	return (EOPNOTSUPP);
   1657 #endif
   1658 }
   1659 /*
   1660  * sysctl helper routine to do kern.lwp.* work.
   1661  */
   1662 static int
   1663 sysctl_kern_lwp(SYSCTLFN_ARGS)
   1664 {
   1665 	struct kinfo_lwp klwp;
   1666 	struct proc *p;
   1667 	struct lwp *l2, *l3;
   1668 	char *where, *dp;
   1669 	int pid, elem_size, elem_count;
   1670 	int buflen, needed, error;
   1671 	bool gotit;
   1672 
   1673 	if (namelen == 1 && name[0] == CTL_QUERY)
   1674 		return (sysctl_query(SYSCTLFN_CALL(rnode)));
   1675 
   1676 	dp = where = oldp;
   1677 	buflen = where != NULL ? *oldlenp : 0;
   1678 	error = needed = 0;
   1679 
   1680 	if (newp != NULL || namelen != 3)
   1681 		return (EINVAL);
   1682 	pid = name[0];
   1683 	elem_size = name[1];
   1684 	elem_count = name[2];
   1685 
   1686 	sysctl_unlock();
   1687 	if (pid == -1) {
   1688 		mutex_enter(&proclist_lock);
   1689 		LIST_FOREACH(p, &allproc, p_list) {
   1690 			/* Grab a hold on the process. */
   1691 			if (!rw_tryenter(&p->p_reflock, RW_READER)) {
   1692 				continue;
   1693 			}
   1694 			mutex_exit(&proclist_lock);
   1695 
   1696 			mutex_enter(&p->p_smutex);
   1697 			LIST_FOREACH(l2, &p->p_lwps, l_sibling) {
   1698 				if (buflen >= elem_size && elem_count > 0) {
   1699 					lwp_lock(l2);
   1700 					fill_lwp(l2, &klwp);
   1701 					lwp_unlock(l2);
   1702 					mutex_exit(&p->p_smutex);
   1703 
   1704 					/*
   1705 					 * Copy out elem_size, but not
   1706 					 * larger than the size of a
   1707 					 * struct kinfo_proc2.
   1708 					 */
   1709 					error = dcopyout(l, &klwp, dp,
   1710 					    min(sizeof(klwp), elem_size));
   1711 					if (error) {
   1712 						rw_exit(&p->p_reflock);
   1713 						goto cleanup;
   1714 					}
   1715 					mutex_enter(&p->p_smutex);
   1716 					LIST_FOREACH(l3, &p->p_lwps,
   1717 					    l_sibling) {
   1718 						if (l2 == l3)
   1719 							break;
   1720 					}
   1721 					if (l3 == NULL) {
   1722 						mutex_exit(&p->p_smutex);
   1723 						rw_exit(&p->p_reflock);
   1724 						error = EAGAIN;
   1725 						goto cleanup;
   1726 					}
   1727 					dp += elem_size;
   1728 					buflen -= elem_size;
   1729 					elem_count--;
   1730 				}
   1731 				needed += elem_size;
   1732 			}
   1733 			mutex_exit(&p->p_smutex);
   1734 
   1735 			/* Drop reference to process. */
   1736 			mutex_enter(&proclist_lock);
   1737 			rw_exit(&p->p_reflock);
   1738 		}
   1739 		mutex_exit(&proclist_lock);
   1740 	} else {
   1741 		mutex_enter(&proclist_lock);
   1742 		p = p_find(pid, PFIND_LOCKED);
   1743 		if (p == NULL) {
   1744 			error = ESRCH;
   1745 			mutex_exit(&proclist_lock);
   1746 			goto cleanup;
   1747 		}
   1748 		/* Grab a hold on the process. */
   1749 		gotit = rw_tryenter(&p->p_reflock, RW_READER);
   1750 		mutex_exit(&proclist_lock);
   1751 		if (!gotit) {
   1752 			error = ESRCH;
   1753 			goto cleanup;
   1754 		}
   1755 
   1756 		mutex_enter(&p->p_smutex);
   1757 		LIST_FOREACH(l2, &p->p_lwps, l_sibling) {
   1758 			if (buflen >= elem_size && elem_count > 0) {
   1759 				lwp_lock(l2);
   1760 				fill_lwp(l2, &klwp);
   1761 				lwp_unlock(l2);
   1762 				mutex_exit(&p->p_smutex);
   1763 				/*
   1764 				 * Copy out elem_size, but not larger than
   1765 				 * the size of a struct kinfo_proc2.
   1766 				 */
   1767 				error = dcopyout(l, &klwp, dp,
   1768 				    min(sizeof(klwp), elem_size));
   1769 				if (error) {
   1770 					rw_exit(&p->p_reflock);
   1771 					goto cleanup;
   1772 				}
   1773 				mutex_enter(&p->p_smutex);
   1774 				LIST_FOREACH(l3, &p->p_lwps, l_sibling) {
   1775 					if (l2 == l3)
   1776 						break;
   1777 				}
   1778 				if (l3 == NULL) {
   1779 					mutex_exit(&p->p_smutex);
   1780 					rw_exit(&p->p_reflock);
   1781 					error = EAGAIN;
   1782 					goto cleanup;
   1783 				}
   1784 				dp += elem_size;
   1785 				buflen -= elem_size;
   1786 				elem_count--;
   1787 			}
   1788 			needed += elem_size;
   1789 		}
   1790 		mutex_exit(&p->p_smutex);
   1791 
   1792 		/* Drop reference to process. */
   1793 		rw_exit(&p->p_reflock);
   1794 	}
   1795 
   1796 	if (where != NULL) {
   1797 		*oldlenp = dp - where;
   1798 		if (needed > *oldlenp) {
   1799 			sysctl_relock();
   1800 			return (ENOMEM);
   1801 		}
   1802 	} else {
   1803 		needed += KERN_LWPSLOP;
   1804 		*oldlenp = needed;
   1805 	}
   1806 	error = 0;
   1807  cleanup:
   1808 	sysctl_relock();
   1809 	return (error);
   1810 }
   1811 
   1812 /*
   1813  * sysctl helper routine for kern.forkfsleep node. Ensures that the
   1814  * given value is not too large or two small, and is at least one
   1815  * timer tick if not zero.
   1816  */
   1817 static int
   1818 sysctl_kern_forkfsleep(SYSCTLFN_ARGS)
   1819 {
   1820 	/* userland sees value in ms, internally is in ticks */
   1821 	extern int forkfsleep;		/* defined in kern/kern_fork.c */
   1822 	int error, timo, lsleep;
   1823 	struct sysctlnode node;
   1824 
   1825 	lsleep = forkfsleep * 1000 / hz;
   1826 	node = *rnode;
   1827 	node.sysctl_data = &lsleep;
   1828 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
   1829 	if (error || newp == NULL)
   1830 		return (error);
   1831 
   1832 	/* refuse negative values, and overly 'long time' */
   1833 	if (lsleep < 0 || lsleep > MAXSLP * 1000)
   1834 		return (EINVAL);
   1835 
   1836 	timo = mstohz(lsleep);
   1837 
   1838 	/* if the interval is >0 ms && <1 tick, use 1 tick */
   1839 	if (lsleep != 0 && timo == 0)
   1840 		forkfsleep = 1;
   1841 	else
   1842 		forkfsleep = timo;
   1843 
   1844 	return (0);
   1845 }
   1846 
   1847 /*
   1848  * sysctl helper routine for kern.root_partition
   1849  */
   1850 static int
   1851 sysctl_kern_root_partition(SYSCTLFN_ARGS)
   1852 {
   1853 	int rootpart = DISKPART(rootdev);
   1854 	struct sysctlnode node = *rnode;
   1855 
   1856 	node.sysctl_data = &rootpart;
   1857 	return (sysctl_lookup(SYSCTLFN_CALL(&node)));
   1858 }
   1859 
   1860 /*
   1861  * sysctl helper function for kern.drivers
   1862  */
   1863 static int
   1864 sysctl_kern_drivers(SYSCTLFN_ARGS)
   1865 {
   1866 	int error;
   1867 	size_t buflen;
   1868 	struct kinfo_drivers kd;
   1869 	char *start, *where;
   1870 	const char *dname;
   1871 	int i;
   1872 	extern struct devsw_conv *devsw_conv;
   1873 	extern int max_devsw_convs;
   1874 	extern kmutex_t devsw_lock;
   1875 
   1876 	if (newp != NULL || namelen != 0)
   1877 		return (EINVAL);
   1878 
   1879 	start = where = oldp;
   1880 	buflen = *oldlenp;
   1881 	if (where == NULL) {
   1882 		*oldlenp = max_devsw_convs * sizeof kd;
   1883 		return 0;
   1884 	}
   1885 
   1886 	/*
   1887 	 * An array of kinfo_drivers structures
   1888 	 */
   1889 	error = 0;
   1890 	sysctl_unlock();
   1891 	mutex_enter(&devsw_lock);
   1892 	for (i = 0; i < max_devsw_convs; i++) {
   1893 		dname = devsw_conv[i].d_name;
   1894 		if (dname == NULL)
   1895 			continue;
   1896 		if (buflen < sizeof kd) {
   1897 			error = ENOMEM;
   1898 			break;
   1899 		}
   1900 		memset(&kd, 0, sizeof(kd));
   1901 		kd.d_bmajor = devsw_conv[i].d_bmajor;
   1902 		kd.d_cmajor = devsw_conv[i].d_cmajor;
   1903 		strlcpy(kd.d_name, dname, sizeof kd.d_name);
   1904 		mutex_exit(&devsw_lock);
   1905 		error = dcopyout(l, &kd, where, sizeof kd);
   1906 		mutex_enter(&devsw_lock);
   1907 		if (error != 0)
   1908 			break;
   1909 		buflen -= sizeof kd;
   1910 		where += sizeof kd;
   1911 	}
   1912 	mutex_exit(&devsw_lock);
   1913 	sysctl_relock();
   1914 	*oldlenp = where - start;
   1915 	return error;
   1916 }
   1917 
   1918 /*
   1919  * sysctl helper function for kern.file2
   1920  */
   1921 static int
   1922 sysctl_kern_file2(SYSCTLFN_ARGS)
   1923 {
   1924 	struct proc *p;
   1925 	struct file *fp, *tp, *np;
   1926 	struct filedesc *fd;
   1927 	struct kinfo_file kf;
   1928 	char *dp;
   1929 	u_int i, op;
   1930 	size_t len, needed, elem_size, out_size;
   1931 	int error, arg, elem_count;
   1932 
   1933 	if (namelen == 1 && name[0] == CTL_QUERY)
   1934 		return (sysctl_query(SYSCTLFN_CALL(rnode)));
   1935 
   1936 	if (namelen != 4)
   1937 		return (EINVAL);
   1938 
   1939 	error = 0;
   1940 	dp = oldp;
   1941 	len = (oldp != NULL) ? *oldlenp : 0;
   1942 	op = name[0];
   1943 	arg = name[1];
   1944 	elem_size = name[2];
   1945 	elem_count = name[3];
   1946 	out_size = MIN(sizeof(kf), elem_size);
   1947 	needed = 0;
   1948 
   1949 	if (elem_size < 1 || elem_count < 0)
   1950 		return (EINVAL);
   1951 
   1952 	switch (op) {
   1953 	case KERN_FILE_BYFILE:
   1954 		/*
   1955 		 * doesn't use arg so it must be zero
   1956 		 */
   1957 		if (arg != 0)
   1958 			return (EINVAL);
   1959 		sysctl_unlock();
   1960 		/*
   1961 		 * allocate dummy file descriptor to make position in list
   1962 		 */
   1963 		if ((tp = fgetdummy()) == NULL) {
   1964 		 	sysctl_relock();
   1965 			return ENOMEM;
   1966 		}
   1967 		mutex_enter(&filelist_lock);
   1968 		for (fp = LIST_FIRST(&filehead); fp != NULL; fp = np) {
   1969 			np = LIST_NEXT(fp, f_list);
   1970 			mutex_enter(&fp->f_lock);
   1971 			if (fp->f_count == 0) {
   1972 				mutex_exit(&fp->f_lock);
   1973 				continue;
   1974 			}
   1975 			/*
   1976 			 * XXX Need to prevent that from being an alternative
   1977 			 * XXX way for getting process information.
   1978 			 */
   1979 			if (kauth_authorize_generic(l->l_cred,
   1980 			    KAUTH_GENERIC_CANSEE, fp->f_cred) != 0) {
   1981 				mutex_exit(&fp->f_lock);
   1982 				continue;
   1983 			}
   1984 			if (len >= elem_size && elem_count > 0) {
   1985 				fill_file(&kf, fp, NULL, 0);
   1986 				LIST_INSERT_AFTER(fp, tp, f_list);
   1987 				mutex_exit(&fp->f_lock);
   1988 				mutex_exit(&filelist_lock);
   1989 				error = dcopyout(l, &kf, dp, out_size);
   1990 				mutex_enter(&filelist_lock);
   1991 				np = LIST_NEXT(tp, f_list);
   1992 				LIST_REMOVE(tp, f_list);
   1993 				if (error) {
   1994 					break;
   1995 				}
   1996 				dp += elem_size;
   1997 				len -= elem_size;
   1998 			} else {
   1999 				mutex_exit(&fp->f_lock);
   2000 			}
   2001 			if (elem_count > 0) {
   2002 				needed += elem_size;
   2003 				if (elem_count != INT_MAX)
   2004 					elem_count--;
   2005 			}
   2006 		}
   2007 		mutex_exit(&filelist_lock);
   2008 		fputdummy(tp);
   2009 		sysctl_relock();
   2010 		break;
   2011 	case KERN_FILE_BYPID:
   2012 		if (arg < -1)
   2013 			/* -1 means all processes */
   2014 			return (EINVAL);
   2015 		sysctl_unlock();
   2016 		mutex_enter(&proclist_lock);
   2017 		LIST_FOREACH(p, &allproc, p_list) {
   2018 			if (p->p_stat == SIDL) {
   2019 				/* skip embryonic processes */
   2020 				continue;
   2021 			}
   2022 			if (arg > 0 && p->p_pid != arg) {
   2023 				/* pick only the one we want */
   2024 				/* XXX want 0 to mean "kernel files" */
   2025 				continue;
   2026 			}
   2027 			mutex_enter(&p->p_mutex);
   2028 			error = kauth_authorize_process(l->l_cred,
   2029 			    KAUTH_PROCESS_CANSEE, p,
   2030 			    KAUTH_ARG(KAUTH_REQ_PROCESS_CANSEE_OPENFILES),
   2031 			    NULL, NULL);
   2032 			mutex_exit(&p->p_mutex);
   2033 			if (error != 0) {
   2034 				continue;
   2035 			}
   2036 
   2037 			/*
   2038 			 * Grab a hold on the process.
   2039 			 */
   2040 			if (!rw_tryenter(&p->p_reflock, RW_READER)) {
   2041 				continue;
   2042 			}
   2043 			mutex_exit(&proclist_lock);
   2044 
   2045 			/* XXX Do we need to check permission per file? */
   2046 			fd = p->p_fd;
   2047 			rw_enter(&fd->fd_lock, RW_READER);
   2048 			for (i = 0; i < fd->fd_nfiles; i++) {
   2049 				fp = fd->fd_ofiles[i];
   2050 				if (fp == NULL) {
   2051 					continue;
   2052 				}
   2053 				mutex_enter(&fp->f_lock);
   2054 				if (!FILE_IS_USABLE(fp)) {
   2055 					mutex_exit(&fp->f_lock);
   2056 					continue;
   2057 				}
   2058 				if (len >= elem_size && elem_count > 0) {
   2059 					fill_file(&kf, fd->fd_ofiles[i], p, i);
   2060 					mutex_exit(&fp->f_lock);
   2061 					rw_exit(&fd->fd_lock);
   2062 					error = dcopyout(l, &kf, dp, out_size);
   2063 					rw_enter(&fd->fd_lock, RW_READER);
   2064 					if (error)
   2065 						break;
   2066 					dp += elem_size;
   2067 					len -= elem_size;
   2068 				} else {
   2069 					mutex_exit(&fp->f_lock);
   2070 				}
   2071 				if (elem_count > 0) {
   2072 					needed += elem_size;
   2073 					if (elem_count != INT_MAX)
   2074 						elem_count--;
   2075 				}
   2076 			}
   2077 			rw_exit(&fd->fd_lock);
   2078 
   2079 			/*
   2080 			 * Release reference to process.
   2081 			 */
   2082 			mutex_enter(&proclist_lock);
   2083 			rw_exit(&p->p_reflock);
   2084 		}
   2085 		mutex_exit(&proclist_lock);
   2086 		sysctl_relock();
   2087 		break;
   2088 	default:
   2089 		return (EINVAL);
   2090 	}
   2091 
   2092 	if (oldp == NULL)
   2093 		needed += KERN_FILESLOP * elem_size;
   2094 	*oldlenp = needed;
   2095 
   2096 	return (error);
   2097 }
   2098 
   2099 static void
   2100 fill_file(struct kinfo_file *kp, const struct file *fp, struct proc *p, int i)
   2101 {
   2102 
   2103 	memset(kp, 0, sizeof(*kp));
   2104 
   2105 	kp->ki_fileaddr =	PTRTOUINT64(fp);
   2106 	kp->ki_flag =		fp->f_flag;
   2107 	kp->ki_iflags =		fp->f_iflags;
   2108 	kp->ki_ftype =		fp->f_type;
   2109 	kp->ki_count =		fp->f_count;
   2110 	kp->ki_msgcount =	fp->f_msgcount;
   2111 	kp->ki_usecount =	fp->f_usecount;
   2112 	kp->ki_fucred =		PTRTOUINT64(fp->f_cred);
   2113 	kp->ki_fuid =		kauth_cred_geteuid(fp->f_cred);
   2114 	kp->ki_fgid =		kauth_cred_getegid(fp->f_cred);
   2115 	kp->ki_fops =		PTRTOUINT64(fp->f_ops);
   2116 	kp->ki_foffset =	fp->f_offset;
   2117 	kp->ki_fdata =		PTRTOUINT64(fp->f_data);
   2118 
   2119 	/* vnode information to glue this file to something */
   2120 	if (fp->f_type == DTYPE_VNODE) {
   2121 		struct vnode *vp = (struct vnode *)fp->f_data;
   2122 
   2123 		kp->ki_vun =	PTRTOUINT64(vp->v_un.vu_socket);
   2124 		kp->ki_vsize =	vp->v_size;
   2125 		kp->ki_vtype =	vp->v_type;
   2126 		kp->ki_vtag =	vp->v_tag;
   2127 		kp->ki_vdata =	PTRTOUINT64(vp->v_data);
   2128 	}
   2129 
   2130 	/* process information when retrieved via KERN_FILE_BYPID */
   2131 	if (p) {
   2132 		KASSERT(rw_lock_held(&p->p_fd->fd_lock));
   2133 
   2134 		kp->ki_pid =		p->p_pid;
   2135 		kp->ki_fd =		i;
   2136 		kp->ki_ofileflags =	p->p_fd->fd_ofileflags[i];
   2137 	}
   2138 }
   2139 
   2140 static int
   2141 sysctl_doeproc(SYSCTLFN_ARGS)
   2142 {
   2143 	struct eproc *eproc;
   2144 	struct kinfo_proc2 *kproc2;
   2145 	struct kinfo_proc *dp;
   2146 	struct proc *p;
   2147 	char *where, *dp2;
   2148 	int type, op, arg;
   2149 	u_int elem_size, elem_count;
   2150 	size_t buflen, needed;
   2151 	bool match;
   2152 	int error;
   2153 
   2154 	if (namelen == 1 && name[0] == CTL_QUERY)
   2155 		return (sysctl_query(SYSCTLFN_CALL(rnode)));
   2156 
   2157 	dp = oldp;
   2158 	dp2 = where = oldp;
   2159 	buflen = where != NULL ? *oldlenp : 0;
   2160 	error = 0;
   2161 	needed = 0;
   2162 	type = rnode->sysctl_num;
   2163 
   2164 	if (type == KERN_PROC) {
   2165 		if (namelen != 2 && !(namelen == 1 && name[0] == KERN_PROC_ALL))
   2166 			return (EINVAL);
   2167 		op = name[0];
   2168 		if (op != KERN_PROC_ALL)
   2169 			arg = name[1];
   2170 		else
   2171 			arg = 0;		/* Quell compiler warning */
   2172 		elem_size = elem_count = 0;	/* Ditto */
   2173 	} else {
   2174 		if (namelen != 4)
   2175 			return (EINVAL);
   2176 		op = name[0];
   2177 		arg = name[1];
   2178 		elem_size = name[2];
   2179 		elem_count = name[3];
   2180 	}
   2181 
   2182 	sysctl_unlock();
   2183 
   2184 	if (type == KERN_PROC) {
   2185 		eproc = kmem_alloc(sizeof(*eproc), KM_SLEEP);
   2186 		kproc2 = NULL;
   2187 	} else {
   2188 		eproc = NULL;
   2189 		kproc2 = kmem_alloc(sizeof(*kproc2), KM_SLEEP);
   2190 	}
   2191 
   2192 	mutex_enter(&proclist_lock);
   2193 	LIST_FOREACH(p, &allproc, p_list) {
   2194 		/*
   2195 		 * Skip embryonic processes.
   2196 		 */
   2197 		if (p->p_stat == SIDL)
   2198 			continue;
   2199 
   2200 		mutex_enter(&p->p_mutex);
   2201 		error = kauth_authorize_process(l->l_cred,
   2202 		    KAUTH_PROCESS_CANSEE, p,
   2203 		    KAUTH_ARG(KAUTH_REQ_PROCESS_CANSEE_ENTRY), NULL, NULL);
   2204 		if (error != 0) {
   2205 			mutex_exit(&p->p_mutex);
   2206 			continue;
   2207 		}
   2208 
   2209 		/*
   2210 		 * TODO - make more efficient (see notes below).
   2211 		 * do by session.
   2212 		 */
   2213 		switch (op) {
   2214 		case KERN_PROC_PID:
   2215 			/* could do this with just a lookup */
   2216 			match = (p->p_pid == (pid_t)arg);
   2217 			break;
   2218 
   2219 		case KERN_PROC_PGRP:
   2220 			/* could do this by traversing pgrp */
   2221 			match = (p->p_pgrp->pg_id == (pid_t)arg);
   2222 			break;
   2223 
   2224 		case KERN_PROC_SESSION:
   2225 			match = (p->p_session->s_sid == (pid_t)arg);
   2226 			break;
   2227 
   2228 		case KERN_PROC_TTY:
   2229 			match = true;
   2230 			if (arg == (int) KERN_PROC_TTY_REVOKE) {
   2231 				if ((p->p_lflag & PL_CONTROLT) == 0 ||
   2232 				    p->p_session->s_ttyp == NULL ||
   2233 				    p->p_session->s_ttyvp != NULL) {
   2234 				    	match = false;
   2235 				}
   2236 			} else if ((p->p_lflag & PL_CONTROLT) == 0 ||
   2237 			    p->p_session->s_ttyp == NULL) {
   2238 				if ((dev_t)arg != KERN_PROC_TTY_NODEV) {
   2239 					match = false;
   2240 				}
   2241 			} else if (p->p_session->s_ttyp->t_dev != (dev_t)arg) {
   2242 				match = false;
   2243 			}
   2244 			break;
   2245 
   2246 		case KERN_PROC_UID:
   2247 			match = (kauth_cred_geteuid(p->p_cred) == (uid_t)arg);
   2248 			break;
   2249 
   2250 		case KERN_PROC_RUID:
   2251 			match = (kauth_cred_getuid(p->p_cred) == (uid_t)arg);
   2252 			break;
   2253 
   2254 		case KERN_PROC_GID:
   2255 			match = (kauth_cred_getegid(p->p_cred) == (uid_t)arg);
   2256 			break;
   2257 
   2258 		case KERN_PROC_RGID:
   2259 			match = (kauth_cred_getgid(p->p_cred) == (uid_t)arg);
   2260 			break;
   2261 
   2262 		case KERN_PROC_ALL:
   2263 			match = true;
   2264 			/* allow everything */
   2265 			break;
   2266 
   2267 		default:
   2268 			error = EINVAL;
   2269 			mutex_exit(&p->p_mutex);
   2270 			goto cleanup;
   2271 		}
   2272 		if (!match) {
   2273 			mutex_exit(&p->p_mutex);
   2274 			continue;
   2275 		}
   2276 
   2277 		/*
   2278 		 * Grab a hold on the process.
   2279 		 */
   2280 		if (!rw_tryenter(&p->p_reflock, RW_READER)) {
   2281 			mutex_exit(&p->p_mutex);
   2282 			continue;
   2283 		}
   2284 
   2285 		if (type == KERN_PROC) {
   2286 			if (buflen >= sizeof(struct kinfo_proc)) {
   2287 				fill_eproc(p, eproc);
   2288 				mutex_exit(&p->p_mutex);
   2289 				mutex_exit(&proclist_lock);
   2290 				error = dcopyout(l, p, &dp->kp_proc,
   2291 				    sizeof(struct proc));
   2292 				mutex_enter(&proclist_lock);
   2293 				if (error) {
   2294 					rw_exit(&p->p_reflock);
   2295 					goto cleanup;
   2296 				}
   2297 				error = dcopyout(l, eproc, &dp->kp_eproc,
   2298 				    sizeof(*eproc));
   2299 				if (error) {
   2300 					rw_exit(&p->p_reflock);
   2301 					goto cleanup;
   2302 				}
   2303 				dp++;
   2304 				buflen -= sizeof(struct kinfo_proc);
   2305 			} else {
   2306 				mutex_exit(&p->p_mutex);
   2307 			}
   2308 			needed += sizeof(struct kinfo_proc);
   2309 		} else { /* KERN_PROC2 */
   2310 			if (buflen >= elem_size && elem_count > 0) {
   2311 				fill_kproc2(p, kproc2);
   2312 				mutex_exit(&p->p_mutex);
   2313 				mutex_exit(&proclist_lock);
   2314 				/*
   2315 				 * Copy out elem_size, but not larger than
   2316 				 * the size of a struct kinfo_proc2.
   2317 				 */
   2318 				error = dcopyout(l, kproc2, dp2,
   2319 				    min(sizeof(*kproc2), elem_size));
   2320 				mutex_enter(&proclist_lock);
   2321 				if (error) {
   2322 					rw_exit(&p->p_reflock);
   2323 					goto cleanup;
   2324 				}
   2325 				dp2 += elem_size;
   2326 				buflen -= elem_size;
   2327 				elem_count--;
   2328 			} else {
   2329 				mutex_exit(&p->p_mutex);
   2330 			}
   2331 			needed += elem_size;
   2332 		}
   2333 
   2334 		/*
   2335 		 * Release reference to process.
   2336 		 */
   2337 		rw_exit(&p->p_reflock);
   2338 	}
   2339 	mutex_exit(&proclist_lock);
   2340 
   2341 	if (where != NULL) {
   2342 		if (type == KERN_PROC)
   2343 			*oldlenp = (char *)dp - where;
   2344 		else
   2345 			*oldlenp = dp2 - where;
   2346 		if (needed > *oldlenp) {
   2347 			error = ENOMEM;
   2348 			goto out;
   2349 		}
   2350 	} else {
   2351 		needed += KERN_PROCSLOP;
   2352 		*oldlenp = needed;
   2353 	}
   2354 	if (kproc2)
   2355 		kmem_free(kproc2, sizeof(*kproc2));
   2356 	if (eproc)
   2357 		kmem_free(eproc, sizeof(*eproc));
   2358 	sysctl_relock();
   2359 	return 0;
   2360  cleanup:
   2361 	mutex_exit(&proclist_lock);
   2362  out:
   2363 	if (kproc2)
   2364 		kmem_free(kproc2, sizeof(*kproc2));
   2365 	if (eproc)
   2366 		kmem_free(eproc, sizeof(*eproc));
   2367 	sysctl_relock();
   2368 	return error;
   2369 }
   2370 
   2371 /*
   2372  * sysctl helper routine for kern.proc_args pseudo-subtree.
   2373  */
   2374 static int
   2375 sysctl_kern_proc_args(SYSCTLFN_ARGS)
   2376 {
   2377 	struct ps_strings pss;
   2378 	struct proc *p;
   2379 	size_t len, i;
   2380 	struct uio auio;
   2381 	struct iovec aiov;
   2382 	pid_t pid;
   2383 	int nargv, type, error, argvlen;
   2384 	char *arg;
   2385 	char **argv = NULL;
   2386 	char *tmp;
   2387 	struct vmspace *vmspace;
   2388 	vaddr_t psstr_addr;
   2389 	vaddr_t offsetn;
   2390 	vaddr_t offsetv;
   2391 
   2392 	if (namelen == 1 && name[0] == CTL_QUERY)
   2393 		return (sysctl_query(SYSCTLFN_CALL(rnode)));
   2394 
   2395 	if (newp != NULL || namelen != 2)
   2396 		return (EINVAL);
   2397 	pid = name[0];
   2398 	type = name[1];
   2399 	argv = NULL;
   2400 	argvlen = 0;
   2401 
   2402 	switch (type) {
   2403 	case KERN_PROC_ARGV:
   2404 	case KERN_PROC_NARGV:
   2405 	case KERN_PROC_ENV:
   2406 	case KERN_PROC_NENV:
   2407 		/* ok */
   2408 		break;
   2409 	default:
   2410 		return (EINVAL);
   2411 	}
   2412 
   2413 	sysctl_unlock();
   2414 
   2415 	/* check pid */
   2416 	mutex_enter(&proclist_lock);
   2417 	if ((p = p_find(pid, PFIND_LOCKED)) == NULL) {
   2418 		error = EINVAL;
   2419 		goto out_locked;
   2420 	}
   2421 	mutex_enter(&p->p_mutex);
   2422 
   2423 	/* Check permission. */
   2424 	if (type == KERN_PROC_ARGV || type == KERN_PROC_NARGV)
   2425 		error = kauth_authorize_process(l->l_cred, KAUTH_PROCESS_CANSEE,
   2426 		    p, KAUTH_ARG(KAUTH_REQ_PROCESS_CANSEE_ARGS), NULL, NULL);
   2427 	else if (type == KERN_PROC_ENV || type == KERN_PROC_NENV)
   2428 		error = kauth_authorize_process(l->l_cred, KAUTH_PROCESS_CANSEE,
   2429 		    p, KAUTH_ARG(KAUTH_REQ_PROCESS_CANSEE_ENV), NULL, NULL);
   2430 	else
   2431 		error = EINVAL; /* XXXGCC */
   2432 	if (error) {
   2433 		mutex_exit(&p->p_mutex);
   2434 		goto out_locked;
   2435 	}
   2436 
   2437 	if (oldp == NULL) {
   2438 		if (type == KERN_PROC_NARGV || type == KERN_PROC_NENV)
   2439 			*oldlenp = sizeof (int);
   2440 		else
   2441 			*oldlenp = ARG_MAX;	/* XXX XXX XXX */
   2442 		error = 0;
   2443 		mutex_exit(&p->p_mutex);
   2444 		goto out_locked;
   2445 	}
   2446 
   2447 	/*
   2448 	 * Zombies don't have a stack, so we can't read their psstrings.
   2449 	 * System processes also don't have a user stack.
   2450 	 */
   2451 	if (P_ZOMBIE(p) || (p->p_flag & PK_SYSTEM) != 0) {
   2452 		error = EINVAL;
   2453 		mutex_exit(&p->p_mutex);
   2454 		goto out_locked;
   2455 	}
   2456 
   2457 	/*
   2458 	 * Lock the process down in memory.
   2459 	 */
   2460 	psstr_addr = (vaddr_t)p->p_psstr;
   2461 	if (type == KERN_PROC_ARGV || type == KERN_PROC_NARGV) {
   2462 		offsetn = p->p_psnargv;
   2463 		offsetv = p->p_psargv;
   2464 	} else {
   2465 		offsetn = p->p_psnenv;
   2466 		offsetv = p->p_psenv;
   2467 	}
   2468 	vmspace = p->p_vmspace;
   2469 	uvmspace_addref(vmspace);
   2470 	mutex_exit(&p->p_mutex);
   2471 	mutex_exit(&proclist_lock);
   2472 
   2473 	/*
   2474 	 * Allocate a temporary buffer to hold the arguments.
   2475 	 */
   2476 	arg = kmem_alloc(PAGE_SIZE, KM_SLEEP);
   2477 
   2478 	/*
   2479 	 * Read in the ps_strings structure.
   2480 	 */
   2481 	aiov.iov_base = &pss;
   2482 	aiov.iov_len = sizeof(pss);
   2483 	auio.uio_iov = &aiov;
   2484 	auio.uio_iovcnt = 1;
   2485 	auio.uio_offset = psstr_addr;
   2486 	auio.uio_resid = sizeof(pss);
   2487 	auio.uio_rw = UIO_READ;
   2488 	UIO_SETUP_SYSSPACE(&auio);
   2489 	error = uvm_io(&vmspace->vm_map, &auio);
   2490 	if (error)
   2491 		goto done;
   2492 
   2493 	memcpy(&nargv, (char *)&pss + offsetn, sizeof(nargv));
   2494 	if (type == KERN_PROC_NARGV || type == KERN_PROC_NENV) {
   2495 		error = dcopyout(l, &nargv, oldp, sizeof(nargv));
   2496 		*oldlenp = sizeof(nargv);
   2497 		goto done;
   2498 	}
   2499 	/*
   2500 	 * Now read the address of the argument vector.
   2501 	 */
   2502 	switch (type) {
   2503 	case KERN_PROC_ARGV:
   2504 		/* FALLTHROUGH */
   2505 	case KERN_PROC_ENV:
   2506 		memcpy(&tmp, (char *)&pss + offsetv, sizeof(tmp));
   2507 		break;
   2508 	default:
   2509 		error = EINVAL;
   2510 		goto done;
   2511 	}
   2512 
   2513 #ifdef COMPAT_NETBSD32
   2514 	if (p->p_flag & PK_32)
   2515 		len = sizeof(netbsd32_charp) * nargv;
   2516 	else
   2517 #endif
   2518 		len = sizeof(char *) * nargv;
   2519 
   2520 	if ((argvlen = len) != 0)
   2521 		argv = kmem_alloc(len, KM_SLEEP);
   2522 
   2523 	aiov.iov_base = argv;
   2524 	aiov.iov_len = len;
   2525 	auio.uio_iov = &aiov;
   2526 	auio.uio_iovcnt = 1;
   2527 	auio.uio_offset = (off_t)(unsigned long)tmp;
   2528 	auio.uio_resid = len;
   2529 	auio.uio_rw = UIO_READ;
   2530 	UIO_SETUP_SYSSPACE(&auio);
   2531 	error = uvm_io(&vmspace->vm_map, &auio);
   2532 	if (error)
   2533 		goto done;
   2534 
   2535 	/*
   2536 	 * Now copy each string.
   2537 	 */
   2538 	len = 0; /* bytes written to user buffer */
   2539 	for (i = 0; i < nargv; i++) {
   2540 		int finished = 0;
   2541 		vaddr_t base;
   2542 		size_t xlen;
   2543 		int j;
   2544 
   2545 #ifdef COMPAT_NETBSD32
   2546 		if (p->p_flag & PK_32) {
   2547 			netbsd32_charp *argv32;
   2548 
   2549 			argv32 = (netbsd32_charp *)argv;
   2550 			base = (vaddr_t)NETBSD32PTR64(argv32[i]);
   2551 		} else
   2552 #endif
   2553 			base = (vaddr_t)argv[i];
   2554 
   2555 		/*
   2556 		 * The program has messed around with its arguments,
   2557 		 * possibly deleting some, and replacing them with
   2558 		 * NULL's. Treat this as the last argument and not
   2559 		 * a failure.
   2560 		 */
   2561 		if (base == 0)
   2562 			break;
   2563 
   2564 		while (!finished) {
   2565 			xlen = PAGE_SIZE - (base & PAGE_MASK);
   2566 
   2567 			aiov.iov_base = arg;
   2568 			aiov.iov_len = PAGE_SIZE;
   2569 			auio.uio_iov = &aiov;
   2570 			auio.uio_iovcnt = 1;
   2571 			auio.uio_offset = base;
   2572 			auio.uio_resid = xlen;
   2573 			auio.uio_rw = UIO_READ;
   2574 			UIO_SETUP_SYSSPACE(&auio);
   2575 			error = uvm_io(&vmspace->vm_map, &auio);
   2576 			if (error)
   2577 				goto done;
   2578 
   2579 			/* Look for the end of the string */
   2580 			for (j = 0; j < xlen; j++) {
   2581 				if (arg[j] == '\0') {
   2582 					xlen = j + 1;
   2583 					finished = 1;
   2584 					break;
   2585 				}
   2586 			}
   2587 
   2588 			/* Check for user buffer overflow */
   2589 			if (len + xlen > *oldlenp) {
   2590 				finished = 1;
   2591 				if (len > *oldlenp)
   2592 					xlen = 0;
   2593 				else
   2594 					xlen = *oldlenp - len;
   2595 			}
   2596 
   2597 			/* Copyout the page */
   2598 			error = dcopyout(l, arg, (char *)oldp + len, xlen);
   2599 			if (error)
   2600 				goto done;
   2601 
   2602 			len += xlen;
   2603 			base += xlen;
   2604 		}
   2605 	}
   2606 	*oldlenp = len;
   2607 
   2608 done:
   2609 	if (argvlen != 0)
   2610 		kmem_free(argv, argvlen);
   2611 	uvmspace_free(vmspace);
   2612 	kmem_free(arg, PAGE_SIZE);
   2613 	sysctl_relock();
   2614 	return error;
   2615 
   2616 out_locked:
   2617 	mutex_exit(&proclist_lock);
   2618 	sysctl_relock();
   2619 	return error;
   2620 }
   2621 
   2622 static int
   2623 sysctl_security_setidcore(SYSCTLFN_ARGS)
   2624 {
   2625 	int newsize, error;
   2626 	struct sysctlnode node;
   2627 
   2628 	node = *rnode;
   2629 	node.sysctl_data = &newsize;
   2630 	newsize = *(int *)rnode->sysctl_data;
   2631 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
   2632 	if (error || newp == NULL)
   2633 		return error;
   2634 
   2635 	if (kauth_authorize_system(l->l_cred, KAUTH_SYSTEM_SETIDCORE,
   2636 	    0, NULL, NULL, NULL))
   2637 		return (EPERM);
   2638 
   2639 	*(int *)rnode->sysctl_data = newsize;
   2640 
   2641 	return 0;
   2642 }
   2643 
   2644 static int
   2645 sysctl_security_setidcorename(SYSCTLFN_ARGS)
   2646 {
   2647 	int error;
   2648 	char *newsetidcorename;
   2649 	struct sysctlnode node;
   2650 
   2651 	newsetidcorename = PNBUF_GET();
   2652 	node = *rnode;
   2653 	node.sysctl_data = newsetidcorename;
   2654 	memcpy(node.sysctl_data, rnode->sysctl_data, MAXPATHLEN);
   2655 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
   2656 	if (error || newp == NULL) {
   2657 		goto out;
   2658 	}
   2659 	if (kauth_authorize_system(l->l_cred, KAUTH_SYSTEM_SETIDCORE,
   2660 	    0, NULL, NULL, NULL)) {
   2661 		error = EPERM;
   2662 		goto out;
   2663 	}
   2664 	if (strlen(newsetidcorename) == 0) {
   2665 		error = EINVAL;
   2666 		goto out;
   2667 	}
   2668 	memcpy(rnode->sysctl_data, node.sysctl_data, MAXPATHLEN);
   2669 out:
   2670 	PNBUF_PUT(newsetidcorename);
   2671 	return error;
   2672 }
   2673 
   2674 /*
   2675  * sysctl helper routine for kern.cp_id node. Maps cpus to their
   2676  * cpuids.
   2677  */
   2678 static int
   2679 sysctl_kern_cpid(SYSCTLFN_ARGS)
   2680 {
   2681 	struct sysctlnode node = *rnode;
   2682 	uint64_t *cp_id = NULL;
   2683 	int error, n = ncpu;
   2684 	struct cpu_info *ci;
   2685 	CPU_INFO_ITERATOR cii;
   2686 
   2687 	/*
   2688 	 * Here you may either retrieve a single cpu id or the whole
   2689 	 * set. The size you get back when probing depends on what
   2690 	 * you ask for.
   2691 	 */
   2692 	switch (namelen) {
   2693 	case 0:
   2694 		node.sysctl_size = n * sizeof(uint64_t);
   2695 		n = -2; /* ALL */
   2696 		break;
   2697 	case 1:
   2698 		if (name[0] < 0 || name[0] >= n)
   2699 			return (ENOENT); /* ENOSUCHPROCESSOR */
   2700 		node.sysctl_size = sizeof(uint64_t);
   2701 		n = name[0];
   2702 		/*
   2703 		 * adjust these so that sysctl_lookup() will be happy
   2704 		 */
   2705 		name++;
   2706 		namelen--;
   2707 		break;
   2708 	default:
   2709 		return (EINVAL);
   2710 	}
   2711 
   2712 	cp_id = kmem_alloc(node.sysctl_size, KM_SLEEP);
   2713 	if (cp_id == NULL)
   2714 		return (ENOMEM);
   2715 	node.sysctl_data = cp_id;
   2716 	memset(cp_id, 0, node.sysctl_size);
   2717 
   2718 	for (CPU_INFO_FOREACH(cii, ci)) {
   2719 		if (n <= 0)
   2720 			cp_id[0] = ci->ci_cpuid;
   2721 		/*
   2722 		 * if a specific processor was requested and we just
   2723 		 * did it, we're done here
   2724 		 */
   2725 		if (n == 0)
   2726 			break;
   2727 		/*
   2728 		 * if doing "all", skip to next cp_id slot for next processor
   2729 		 */
   2730 		if (n == -2)
   2731 			cp_id++;
   2732 		/*
   2733 		 * if we're doing a specific processor, we're one
   2734 		 * processor closer
   2735 		 */
   2736 		if (n > 0)
   2737 			n--;
   2738 	}
   2739 
   2740 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
   2741 	kmem_free(node.sysctl_data, node.sysctl_size);
   2742 	return (error);
   2743 }
   2744 
   2745 /*
   2746  * sysctl helper routine for hw.usermem and hw.usermem64. Values are
   2747  * calculate on the fly taking into account integer overflow and the
   2748  * current wired count.
   2749  */
   2750 static int
   2751 sysctl_hw_usermem(SYSCTLFN_ARGS)
   2752 {
   2753 	u_int ui;
   2754 	u_quad_t uq;
   2755 	struct sysctlnode node;
   2756 
   2757 	node = *rnode;
   2758 	switch (rnode->sysctl_num) {
   2759 	    case HW_USERMEM:
   2760 		if ((ui = physmem - uvmexp.wired) > (UINT_MAX / PAGE_SIZE))
   2761 			ui = UINT_MAX;
   2762 		else
   2763 			ui *= PAGE_SIZE;
   2764 		node.sysctl_data = &ui;
   2765 		break;
   2766 	case HW_USERMEM64:
   2767 		uq = (u_quad_t)(physmem - uvmexp.wired) * PAGE_SIZE;
   2768 		node.sysctl_data = &uq;
   2769 		break;
   2770 	default:
   2771 		return (EINVAL);
   2772 	}
   2773 
   2774 	return (sysctl_lookup(SYSCTLFN_CALL(&node)));
   2775 }
   2776 
   2777 /*
   2778  * sysctl helper routine for kern.cnmagic node. Pulls the old value
   2779  * out, encoded, and stuffs the new value in for decoding.
   2780  */
   2781 static int
   2782 sysctl_hw_cnmagic(SYSCTLFN_ARGS)
   2783 {
   2784 	char magic[CNS_LEN];
   2785 	int error;
   2786 	struct sysctlnode node;
   2787 
   2788 	if (oldp)
   2789 		cn_get_magic(magic, CNS_LEN);
   2790 	node = *rnode;
   2791 	node.sysctl_data = &magic[0];
   2792 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
   2793 	if (error || newp == NULL)
   2794 		return (error);
   2795 
   2796 	return (cn_set_magic(magic));
   2797 }
   2798 
   2799 /*
   2800  * ********************************************************************
   2801  * section 3: public helper routines that are used for more than one
   2802  * node
   2803  * ********************************************************************
   2804  */
   2805 
   2806 /*
   2807  * sysctl helper routine for the kern.root_device node and some ports'
   2808  * machdep.root_device nodes.
   2809  */
   2810 int
   2811 sysctl_root_device(SYSCTLFN_ARGS)
   2812 {
   2813 	struct sysctlnode node;
   2814 
   2815 	node = *rnode;
   2816 	node.sysctl_data = root_device->dv_xname;
   2817 	node.sysctl_size = strlen(root_device->dv_xname) + 1;
   2818 	return (sysctl_lookup(SYSCTLFN_CALL(&node)));
   2819 }
   2820 
   2821 /*
   2822  * sysctl helper routine for kern.consdev, dependent on the current
   2823  * state of the console. Also used for machdep.console_device on some
   2824  * ports.
   2825  */
   2826 int
   2827 sysctl_consdev(SYSCTLFN_ARGS)
   2828 {
   2829 	dev_t consdev;
   2830 	struct sysctlnode node;
   2831 
   2832 	if (cn_tab != NULL)
   2833 		consdev = cn_tab->cn_dev;
   2834 	else
   2835 		consdev = NODEV;
   2836 	node = *rnode;
   2837 	node.sysctl_data = &consdev;
   2838 	node.sysctl_size = sizeof(consdev);
   2839 	return (sysctl_lookup(SYSCTLFN_CALL(&node)));
   2840 }
   2841 
   2842 /*
   2843  * ********************************************************************
   2844  * section 4: support for some helpers
   2845  * ********************************************************************
   2846  */
   2847 
   2848 /*
   2849  * Fill in a kinfo_proc2 structure for the specified process.
   2850  */
   2851 static void
   2852 fill_kproc2(struct proc *p, struct kinfo_proc2 *ki)
   2853 {
   2854 	struct tty *tp;
   2855 	struct lwp *l, *l2;
   2856 	struct timeval ut, st, rt;
   2857 	sigset_t ss1, ss2;
   2858 
   2859 	KASSERT(mutex_owned(&proclist_lock));
   2860 	KASSERT(mutex_owned(&p->p_mutex));
   2861 
   2862 	memset(ki, 0, sizeof(*ki));
   2863 
   2864 	ki->p_paddr = PTRTOUINT64(p);
   2865 	ki->p_fd = PTRTOUINT64(p->p_fd);
   2866 	ki->p_cwdi = PTRTOUINT64(p->p_cwdi);
   2867 	ki->p_stats = PTRTOUINT64(p->p_stats);
   2868 	ki->p_limit = PTRTOUINT64(p->p_limit);
   2869 	ki->p_vmspace = PTRTOUINT64(p->p_vmspace);
   2870 	ki->p_sigacts = PTRTOUINT64(p->p_sigacts);
   2871 	ki->p_sess = PTRTOUINT64(p->p_session);
   2872 	ki->p_tsess = 0;	/* may be changed if controlling tty below */
   2873 	ki->p_ru = PTRTOUINT64(&p->p_stats->p_ru);
   2874 
   2875 	ki->p_eflag = 0;
   2876 	ki->p_exitsig = p->p_exitsig;
   2877 
   2878 	ki->p_flag = sysctl_map_flags(sysctl_flagmap, p->p_flag);
   2879 	ki->p_flag |= sysctl_map_flags(sysctl_sflagmap, p->p_sflag);
   2880 	ki->p_flag |= sysctl_map_flags(sysctl_slflagmap, p->p_slflag);
   2881 	ki->p_flag |= sysctl_map_flags(sysctl_lflagmap, p->p_lflag);
   2882 	ki->p_flag |= sysctl_map_flags(sysctl_stflagmap, p->p_stflag);
   2883 
   2884 	ki->p_pid = p->p_pid;
   2885 	if (p->p_pptr)
   2886 		ki->p_ppid = p->p_pptr->p_pid;
   2887 	else
   2888 		ki->p_ppid = 0;
   2889 	ki->p_sid = p->p_session->s_sid;
   2890 	ki->p__pgid = p->p_pgrp->pg_id;
   2891 
   2892 	ki->p_tpgid = NO_PGID;	/* may be changed if controlling tty below */
   2893 
   2894 	ki->p_uid = kauth_cred_geteuid(p->p_cred);
   2895 	ki->p_ruid = kauth_cred_getuid(p->p_cred);
   2896 	ki->p_gid = kauth_cred_getegid(p->p_cred);
   2897 	ki->p_rgid = kauth_cred_getgid(p->p_cred);
   2898 	ki->p_svuid = kauth_cred_getsvuid(p->p_cred);
   2899 	ki->p_svgid = kauth_cred_getsvgid(p->p_cred);
   2900 
   2901 	ki->p_ngroups = kauth_cred_ngroups(p->p_cred);
   2902 	kauth_cred_getgroups(p->p_cred, ki->p_groups,
   2903 	    min(ki->p_ngroups, sizeof(ki->p_groups) / sizeof(ki->p_groups[0])),
   2904 	    UIO_SYSSPACE);
   2905 
   2906 	ki->p_jobc = p->p_pgrp->pg_jobc;
   2907 	if ((p->p_lflag & PL_CONTROLT) && (tp = p->p_session->s_ttyp)) {
   2908 		ki->p_tdev = tp->t_dev;
   2909 		ki->p_tpgid = tp->t_pgrp ? tp->t_pgrp->pg_id : NO_PGID;
   2910 		ki->p_tsess = PTRTOUINT64(tp->t_session);
   2911 	} else {
   2912 		ki->p_tdev = NODEV;
   2913 	}
   2914 
   2915 	mutex_enter(&p->p_smutex);
   2916 
   2917 	ki->p_uticks = p->p_uticks;
   2918 	ki->p_sticks = p->p_sticks;
   2919 	ki->p_iticks = p->p_iticks;
   2920 
   2921 	ki->p_tracep = PTRTOUINT64(p->p_tracep);
   2922 	ki->p_traceflag = p->p_traceflag;
   2923 
   2924 	memcpy(&ki->p_sigignore, &p->p_sigctx.ps_sigignore,sizeof(ki_sigset_t));
   2925 	memcpy(&ki->p_sigcatch, &p->p_sigctx.ps_sigcatch, sizeof(ki_sigset_t));
   2926 
   2927 	ki->p_cpticks = 0;
   2928 	ki->p_pctcpu = p->p_pctcpu;
   2929 	ki->p_estcpu = 0;
   2930 	ss1 = p->p_sigpend.sp_set;
   2931 	LIST_FOREACH(l, &p->p_lwps, l_sibling) {
   2932 		/* This is hardly correct, but... */
   2933 		sigplusset(&l->l_sigpend.sp_set, &ss1);
   2934 		sigplusset(&l->l_sigmask, &ss2);
   2935 		ki->p_cpticks += l->l_cpticks;
   2936 		ki->p_pctcpu += l->l_pctcpu;
   2937 		ki->p_estcpu += l->l_estcpu;
   2938 	}
   2939 	memcpy(&ki->p_siglist, &ss1, sizeof(ki_sigset_t));
   2940 	memcpy(&ki->p_sigmask, &ss2, sizeof(ki_sigset_t));
   2941 
   2942 	ki->p_stat = p->p_stat; /* Will likely be overridden by LWP status */
   2943 	ki->p_realstat = p->p_stat;
   2944 	ki->p_nice = p->p_nice;
   2945 
   2946 	ki->p_xstat = p->p_xstat;
   2947 	ki->p_acflag = p->p_acflag;
   2948 
   2949 	strncpy(ki->p_comm, p->p_comm,
   2950 	    min(sizeof(ki->p_comm), sizeof(p->p_comm)));
   2951 
   2952 	strncpy(ki->p_login, p->p_session->s_login,
   2953 	    min(sizeof ki->p_login - 1, sizeof p->p_session->s_login));
   2954 
   2955 	ki->p_nlwps = p->p_nlwps;
   2956 	ki->p_realflag = ki->p_flag;
   2957 
   2958 	if (p->p_stat == SIDL || P_ZOMBIE(p)) {
   2959 		ki->p_vm_rssize = 0;
   2960 		ki->p_vm_tsize = 0;
   2961 		ki->p_vm_dsize = 0;
   2962 		ki->p_vm_ssize = 0;
   2963 		ki->p_nrlwps = 0;
   2964 		l = NULL;
   2965 	} else {
   2966 		struct vmspace *vm = p->p_vmspace;
   2967 		int tmp;
   2968 
   2969 		ki->p_vm_rssize = vm_resident_count(vm);
   2970 		ki->p_vm_tsize = vm->vm_tsize;
   2971 		ki->p_vm_dsize = vm->vm_dsize;
   2972 		ki->p_vm_ssize = vm->vm_ssize;
   2973 
   2974 		/* Pick a "representative" LWP */
   2975 		l = proc_representative_lwp(p, &tmp, 1);
   2976 		lwp_lock(l);
   2977 		ki->p_nrlwps = tmp;
   2978 		ki->p_forw = 0;
   2979 		ki->p_back = 0;
   2980 		ki->p_addr = PTRTOUINT64(l->l_addr);
   2981 		ki->p_stat = l->l_stat;
   2982 		ki->p_flag |= sysctl_map_flags(sysctl_lwpflagmap, l->l_flag);
   2983 		ki->p_swtime = l->l_swtime;
   2984 		ki->p_slptime = l->l_slptime;
   2985 		if (l->l_stat == LSONPROC)
   2986 			ki->p_schedflags = l->l_cpu->ci_schedstate.spc_flags;
   2987 		else
   2988 			ki->p_schedflags = 0;
   2989 		ki->p_holdcnt = l->l_holdcnt;
   2990 		ki->p_priority = lwp_eprio(l);
   2991 		ki->p_usrpri = l->l_priority;
   2992 		if (l->l_wmesg)
   2993 			strncpy(ki->p_wmesg, l->l_wmesg, sizeof(ki->p_wmesg));
   2994 		ki->p_wchan = PTRTOUINT64(l->l_wchan);
   2995 		lwp_unlock(l);
   2996 	}
   2997 	if (p->p_session->s_ttyvp)
   2998 		ki->p_eflag |= EPROC_CTTY;
   2999 	if (SESS_LEADER(p))
   3000 		ki->p_eflag |= EPROC_SLEADER;
   3001 
   3002 	/* XXX Is this double check necessary? */
   3003 	if (P_ZOMBIE(p)) {
   3004 		ki->p_uvalid = 0;
   3005 		ki->p_rtime_sec = 0;
   3006 		ki->p_rtime_usec = 0;
   3007 	} else {
   3008 		ki->p_uvalid = 1;
   3009 
   3010 		ki->p_ustart_sec = p->p_stats->p_start.tv_sec;
   3011 		ki->p_ustart_usec = p->p_stats->p_start.tv_usec;
   3012 
   3013 		calcru(p, &ut, &st, NULL, &rt);
   3014 		ki->p_rtime_sec = rt.tv_sec;
   3015 		ki->p_rtime_usec = rt.tv_usec;
   3016 		ki->p_uutime_sec = ut.tv_sec;
   3017 		ki->p_uutime_usec = ut.tv_usec;
   3018 		ki->p_ustime_sec = st.tv_sec;
   3019 		ki->p_ustime_usec = st.tv_usec;
   3020 
   3021 		ki->p_uru_maxrss = p->p_stats->p_ru.ru_maxrss;
   3022 		ki->p_uru_ixrss = p->p_stats->p_ru.ru_ixrss;
   3023 		ki->p_uru_idrss = p->p_stats->p_ru.ru_idrss;
   3024 		ki->p_uru_isrss = p->p_stats->p_ru.ru_isrss;
   3025 		ki->p_uru_minflt = p->p_stats->p_ru.ru_minflt;
   3026 		ki->p_uru_majflt = p->p_stats->p_ru.ru_majflt;
   3027 		ki->p_uru_nswap = p->p_stats->p_ru.ru_nswap;
   3028 		ki->p_uru_inblock = p->p_stats->p_ru.ru_inblock;
   3029 		ki->p_uru_oublock = p->p_stats->p_ru.ru_oublock;
   3030 		ki->p_uru_msgsnd = p->p_stats->p_ru.ru_msgsnd;
   3031 		ki->p_uru_msgrcv = p->p_stats->p_ru.ru_msgrcv;
   3032 		ki->p_uru_nsignals = p->p_stats->p_ru.ru_nsignals;
   3033 
   3034 		ki->p_uru_nvcsw = 0;
   3035 		ki->p_uru_nivcsw = 0;
   3036 		LIST_FOREACH(l2, &p->p_lwps, l_sibling) {
   3037 			ki->p_uru_nvcsw += (l->l_ncsw - l->l_nivcsw);
   3038 			ki->p_uru_nivcsw += l->l_nivcsw;
   3039 		}
   3040 
   3041 		timeradd(&p->p_stats->p_cru.ru_utime,
   3042 			 &p->p_stats->p_cru.ru_stime, &ut);
   3043 		ki->p_uctime_sec = ut.tv_sec;
   3044 		ki->p_uctime_usec = ut.tv_usec;
   3045 	}
   3046 	if (l != NULL)
   3047 		ki->p_cpuid = l->l_cpu->ci_cpuid;
   3048 
   3049 	mutex_exit(&p->p_smutex);
   3050 }
   3051 
   3052 /*
   3053  * Fill in a kinfo_lwp structure for the specified lwp.
   3054  */
   3055 static void
   3056 fill_lwp(struct lwp *l, struct kinfo_lwp *kl)
   3057 {
   3058 	struct proc *p = l->l_proc;
   3059 	struct timeval tv;
   3060 
   3061 	kl->l_forw = 0;
   3062 	kl->l_back = 0;
   3063 	kl->l_laddr = PTRTOUINT64(l);
   3064 	kl->l_addr = PTRTOUINT64(l->l_addr);
   3065 	kl->l_stat = l->l_stat;
   3066 	kl->l_lid = l->l_lid;
   3067 	kl->l_flag = sysctl_map_flags(sysctl_lwpprflagmap, l->l_prflag);
   3068 
   3069 	kl->l_swtime = l->l_swtime;
   3070 	kl->l_slptime = l->l_slptime;
   3071 	if (l->l_stat == LSONPROC)
   3072 		kl->l_schedflags = l->l_cpu->ci_schedstate.spc_flags;
   3073 	else
   3074 		kl->l_schedflags = 0;
   3075 	kl->l_holdcnt = l->l_holdcnt;
   3076 	kl->l_priority = lwp_eprio(l);
   3077 	kl->l_usrpri = l->l_priority;
   3078 	if (l->l_wmesg)
   3079 		strncpy(kl->l_wmesg, l->l_wmesg, sizeof(kl->l_wmesg));
   3080 	kl->l_wchan = PTRTOUINT64(l->l_wchan);
   3081 	kl->l_cpuid = l->l_cpu->ci_cpuid;
   3082 	bintime2timeval(&l->l_rtime, &tv);
   3083 	kl->l_rtime_sec = tv.tv_sec;
   3084 	kl->l_rtime_usec = tv.tv_usec;
   3085 	kl->l_cpticks = l->l_cpticks;
   3086 	kl->l_pctcpu = l->l_pctcpu;
   3087 	kl->l_pid = p->p_pid;
   3088 	if (l->l_name == NULL)
   3089 		kl->l_name[0] = '\0';
   3090 	else
   3091 		strlcpy(kl->l_name, l->l_name, sizeof(kl->l_name));
   3092 }
   3093 
   3094 /*
   3095  * Fill in an eproc structure for the specified process.
   3096  */
   3097 void
   3098 fill_eproc(struct proc *p, struct eproc *ep)
   3099 {
   3100 	struct tty *tp;
   3101 	struct lwp *l;
   3102 
   3103 	KASSERT(mutex_owned(&proclist_lock));
   3104 	KASSERT(mutex_owned(&p->p_mutex));
   3105 
   3106 	ep->e_paddr = p;
   3107 	ep->e_sess = p->p_session;
   3108 	kauth_cred_topcred(p->p_cred, &ep->e_pcred);
   3109 	kauth_cred_toucred(p->p_cred, &ep->e_ucred);
   3110 	if (p->p_stat == SIDL || P_ZOMBIE(p)) {
   3111 		ep->e_vm.vm_rssize = 0;
   3112 		ep->e_vm.vm_tsize = 0;
   3113 		ep->e_vm.vm_dsize = 0;
   3114 		ep->e_vm.vm_ssize = 0;
   3115 		/* ep->e_vm.vm_pmap = XXX; */
   3116 	} else {
   3117 		struct vmspace *vm = p->p_vmspace;
   3118 
   3119 		ep->e_vm.vm_rssize = vm_resident_count(vm);
   3120 		ep->e_vm.vm_tsize = vm->vm_tsize;
   3121 		ep->e_vm.vm_dsize = vm->vm_dsize;
   3122 		ep->e_vm.vm_ssize = vm->vm_ssize;
   3123 
   3124 		/* Pick a "representative" LWP */
   3125 		mutex_enter(&p->p_smutex);
   3126 		l = proc_representative_lwp(p, NULL, 1);
   3127 		lwp_lock(l);
   3128 		if (l->l_wmesg)
   3129 			strncpy(ep->e_wmesg, l->l_wmesg, WMESGLEN);
   3130 		lwp_unlock(l);
   3131 		mutex_exit(&p->p_smutex);
   3132 	}
   3133 	if (p->p_pptr)
   3134 		ep->e_ppid = p->p_pptr->p_pid;
   3135 	else
   3136 		ep->e_ppid = 0;
   3137 	ep->e_pgid = p->p_pgrp->pg_id;
   3138 	ep->e_sid = ep->e_sess->s_sid;
   3139 	ep->e_jobc = p->p_pgrp->pg_jobc;
   3140 	if ((p->p_lflag & PL_CONTROLT) &&
   3141 	    (tp = ep->e_sess->s_ttyp)) {
   3142 		ep->e_tdev = tp->t_dev;
   3143 		ep->e_tpgid = tp->t_pgrp ? tp->t_pgrp->pg_id : NO_PGID;
   3144 		ep->e_tsess = tp->t_session;
   3145 	} else
   3146 		ep->e_tdev = NODEV;
   3147 
   3148 	ep->e_xsize = ep->e_xrssize = 0;
   3149 	ep->e_xccount = ep->e_xswrss = 0;
   3150 	ep->e_flag = ep->e_sess->s_ttyvp ? EPROC_CTTY : 0;
   3151 	if (SESS_LEADER(p))
   3152 		ep->e_flag |= EPROC_SLEADER;
   3153 	strncpy(ep->e_login, ep->e_sess->s_login, MAXLOGNAME);
   3154 }
   3155 
   3156 u_int
   3157 sysctl_map_flags(const u_int *map, u_int word)
   3158 {
   3159 	u_int rv;
   3160 
   3161 	for (rv = 0; *map != 0; map += 2)
   3162 		if ((word & map[0]) != 0)
   3163 			rv |= map[1];
   3164 
   3165 	return rv;
   3166 }
   3167