xref: /freebsd/sys/kern/subr_param.c (revision fd5e3f3ec6c6248e892c9e7b2f17da3bfe7b6837)
1 /*-
2  * Copyright (c) 1980, 1986, 1989, 1993
3  *	The Regents of the University of California.  All rights reserved.
4  * (c) UNIX System Laboratories, Inc.
5  * All or some portions of this file are derived from material licensed
6  * to the University of California by American Telephone and Telegraph
7  * Co. or Unix System Laboratories, Inc. and are reproduced herein with
8  * the permission of UNIX System Laboratories, Inc.
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. Neither the name of the University nor the names of its contributors
19  *    may be used to endorse or promote products derived from this software
20  *    without specific prior written permission.
21  *
22  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
23  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
24  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
25  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
26  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
27  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
28  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
29  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
30  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
31  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
32  * SUCH DAMAGE.
33  *
34  *	@(#)param.c	8.3 (Berkeley) 8/20/94
35  */
36 
37 #include <sys/cdefs.h>
38 __FBSDID("$FreeBSD$");
39 
40 #include "opt_param.h"
41 #include "opt_msgbuf.h"
42 #include "opt_maxusers.h"
43 
44 #include <sys/param.h>
45 #include <sys/systm.h>
46 #include <sys/kernel.h>
47 #include <sys/limits.h>
48 #include <sys/msgbuf.h>
49 #include <sys/sysctl.h>
50 #include <sys/proc.h>
51 
52 #include <vm/vm.h>
53 #include <vm/vm_param.h>
54 #include <vm/pmap.h>
55 
56 /*
57  * System parameter formulae.
58  */
59 
60 #ifndef HZ
61 #  if defined(__mips__) || defined(__arm__)
62 #    define	HZ 100
63 #  else
64 #    define	HZ 1000
65 #  endif
66 #  ifndef HZ_VM
67 #    define	HZ_VM 100
68 #  endif
69 #else
70 #  ifndef HZ_VM
71 #    define	HZ_VM HZ
72 #  endif
73 #endif
74 #define	NPROC (20 + 16 * maxusers)
75 #ifndef NBUF
76 #define NBUF 0
77 #endif
78 #ifndef MAXFILES
79 #define	MAXFILES (40 + 32 * maxusers)
80 #endif
81 
82 static int sysctl_kern_vm_guest(SYSCTL_HANDLER_ARGS);
83 
84 int	hz;				/* system clock's frequency */
85 int	tick;				/* usec per tick (1000000 / hz) */
86 struct bintime tick_bt;			/* bintime per tick (1s / hz) */
87 sbintime_t tick_sbt;
88 int	maxusers;			/* base tunable */
89 int	maxproc;			/* maximum # of processes */
90 int	maxprocperuid;			/* max # of procs per user */
91 int	maxfiles;			/* sys. wide open files limit */
92 int	maxfilesperproc;		/* per-proc open files limit */
93 int	msgbufsize;			/* size of kernel message buffer */
94 int	nbuf;
95 int	bio_transient_maxcnt;
96 int	ngroups_max;			/* max # groups per process */
97 int	nswbuf;
98 pid_t	pid_max = PID_MAX;
99 long	maxswzone;			/* max swmeta KVA storage */
100 long	maxbcache;			/* max buffer cache KVA storage */
101 long	maxpipekva;			/* Limit on pipe KVA */
102 int	vm_guest = VM_GUEST_NO;		/* Running as virtual machine guest? */
103 u_long	maxtsiz;			/* max text size */
104 u_long	dfldsiz;			/* initial data size limit */
105 u_long	maxdsiz;			/* max data size */
106 u_long	dflssiz;			/* initial stack size limit */
107 u_long	maxssiz;			/* max stack size */
108 u_long	sgrowsiz;			/* amount to grow stack */
109 
110 SYSCTL_INT(_kern, OID_AUTO, hz, CTLFLAG_RDTUN | CTLFLAG_NOFETCH, &hz, 0,
111     "Number of clock ticks per second");
112 SYSCTL_INT(_kern, OID_AUTO, nbuf, CTLFLAG_RDTUN | CTLFLAG_NOFETCH, &nbuf, 0,
113     "Number of buffers in the buffer cache");
114 SYSCTL_INT(_kern, OID_AUTO, nswbuf, CTLFLAG_RDTUN | CTLFLAG_NOFETCH, &nswbuf, 0,
115     "Number of swap buffers");
116 SYSCTL_INT(_kern, OID_AUTO, msgbufsize, CTLFLAG_RDTUN | CTLFLAG_NOFETCH, &msgbufsize, 0,
117     "Size of the kernel message buffer");
118 SYSCTL_LONG(_kern, OID_AUTO, maxswzone, CTLFLAG_RDTUN | CTLFLAG_NOFETCH, &maxswzone, 0,
119     "Maximum memory for swap metadata");
120 SYSCTL_LONG(_kern, OID_AUTO, maxbcache, CTLFLAG_RDTUN | CTLFLAG_NOFETCH, &maxbcache, 0,
121     "Maximum value of vfs.maxbufspace");
122 SYSCTL_INT(_kern, OID_AUTO, bio_transient_maxcnt, CTLFLAG_RDTUN | CTLFLAG_NOFETCH,
123     &bio_transient_maxcnt, 0,
124     "Maximum number of transient BIOs mappings");
125 SYSCTL_ULONG(_kern, OID_AUTO, maxtsiz, CTLFLAG_RWTUN | CTLFLAG_NOFETCH, &maxtsiz, 0,
126     "Maximum text size");
127 SYSCTL_ULONG(_kern, OID_AUTO, dfldsiz, CTLFLAG_RWTUN | CTLFLAG_NOFETCH, &dfldsiz, 0,
128     "Initial data size limit");
129 SYSCTL_ULONG(_kern, OID_AUTO, maxdsiz, CTLFLAG_RWTUN | CTLFLAG_NOFETCH, &maxdsiz, 0,
130     "Maximum data size");
131 SYSCTL_ULONG(_kern, OID_AUTO, dflssiz, CTLFLAG_RWTUN | CTLFLAG_NOFETCH, &dflssiz, 0,
132     "Initial stack size limit");
133 SYSCTL_ULONG(_kern, OID_AUTO, maxssiz, CTLFLAG_RWTUN | CTLFLAG_NOFETCH, &maxssiz, 0,
134     "Maximum stack size");
135 SYSCTL_ULONG(_kern, OID_AUTO, sgrowsiz, CTLFLAG_RWTUN | CTLFLAG_NOFETCH, &sgrowsiz, 0,
136     "Amount to grow stack on a stack fault");
137 SYSCTL_PROC(_kern, OID_AUTO, vm_guest, CTLFLAG_RD | CTLTYPE_STRING,
138     NULL, 0, sysctl_kern_vm_guest, "A",
139     "Virtual machine guest detected?");
140 
141 /*
142  * The elements of this array are ordered based upon the values of the
143  * corresponding enum VM_GUEST members.
144  */
145 static const char *const vm_guest_sysctl_names[] = {
146 	"none",
147 	"generic",
148 	"xen",
149 	"hv",
150 	"vmware",
151 	"kvm",
152 	"bhyve",
153 	NULL
154 };
155 CTASSERT(nitems(vm_guest_sysctl_names) - 1 == VM_LAST);
156 
157 /*
158  * Boot time overrides that are not scaled against main memory
159  */
160 void
161 init_param1(void)
162 {
163 
164 #if !defined(__mips__) && !defined(__arm64__) && !defined(__sparc64__)
165 	TUNABLE_INT_FETCH("kern.kstack_pages", &kstack_pages);
166 #endif
167 	hz = -1;
168 	TUNABLE_INT_FETCH("kern.hz", &hz);
169 	if (hz == -1)
170 		hz = vm_guest > VM_GUEST_NO ? HZ_VM : HZ;
171 	tick = 1000000 / hz;
172 	tick_sbt = SBT_1S / hz;
173 	tick_bt = sbttobt(tick_sbt);
174 
175 	/*
176 	 * Arrange for ticks to wrap 10 minutes after boot to help catch
177 	 * sign problems sooner.
178 	 */
179 	ticks = INT_MAX - (hz * 10 * 60);
180 
181 #ifdef VM_SWZONE_SIZE_MAX
182 	maxswzone = VM_SWZONE_SIZE_MAX;
183 #endif
184 	TUNABLE_LONG_FETCH("kern.maxswzone", &maxswzone);
185 #ifdef VM_BCACHE_SIZE_MAX
186 	maxbcache = VM_BCACHE_SIZE_MAX;
187 #endif
188 	TUNABLE_LONG_FETCH("kern.maxbcache", &maxbcache);
189 	msgbufsize = MSGBUF_SIZE;
190 	TUNABLE_INT_FETCH("kern.msgbufsize", &msgbufsize);
191 
192 	maxtsiz = MAXTSIZ;
193 	TUNABLE_ULONG_FETCH("kern.maxtsiz", &maxtsiz);
194 	dfldsiz = DFLDSIZ;
195 	TUNABLE_ULONG_FETCH("kern.dfldsiz", &dfldsiz);
196 	maxdsiz = MAXDSIZ;
197 	TUNABLE_ULONG_FETCH("kern.maxdsiz", &maxdsiz);
198 	dflssiz = DFLSSIZ;
199 	TUNABLE_ULONG_FETCH("kern.dflssiz", &dflssiz);
200 	maxssiz = MAXSSIZ;
201 	TUNABLE_ULONG_FETCH("kern.maxssiz", &maxssiz);
202 	sgrowsiz = SGROWSIZ;
203 	TUNABLE_ULONG_FETCH("kern.sgrowsiz", &sgrowsiz);
204 
205 	/*
206 	 * Let the administrator set {NGROUPS_MAX}, but disallow values
207 	 * less than NGROUPS_MAX which would violate POSIX.1-2008 or
208 	 * greater than INT_MAX-1 which would result in overflow.
209 	 */
210 	ngroups_max = NGROUPS_MAX;
211 	TUNABLE_INT_FETCH("kern.ngroups", &ngroups_max);
212 	if (ngroups_max < NGROUPS_MAX)
213 		ngroups_max = NGROUPS_MAX;
214 
215 	/*
216 	 * Only allow to lower the maximal pid.
217 	 * Prevent setting up a non-bootable system if pid_max is too low.
218 	 */
219 	TUNABLE_INT_FETCH("kern.pid_max", &pid_max);
220 	if (pid_max > PID_MAX)
221 		pid_max = PID_MAX;
222 	else if (pid_max < 300)
223 		pid_max = 300;
224 
225 	TUNABLE_INT_FETCH("vfs.unmapped_buf_allowed", &unmapped_buf_allowed);
226 }
227 
228 /*
229  * Boot time overrides that are scaled against main memory
230  */
231 void
232 init_param2(long physpages)
233 {
234 
235 	/* Base parameters */
236 	maxusers = MAXUSERS;
237 	TUNABLE_INT_FETCH("kern.maxusers", &maxusers);
238 	if (maxusers == 0) {
239 		maxusers = physpages / (2 * 1024 * 1024 / PAGE_SIZE);
240 		if (maxusers < 32)
241 			maxusers = 32;
242 #ifdef VM_MAX_AUTOTUNE_MAXUSERS
243                 if (maxusers > VM_MAX_AUTOTUNE_MAXUSERS)
244                         maxusers = VM_MAX_AUTOTUNE_MAXUSERS;
245 #endif
246                 /*
247                  * Scales down the function in which maxusers grows once
248                  * we hit 384.
249                  */
250                 if (maxusers > 384)
251                         maxusers = 384 + ((maxusers - 384) / 8);
252         }
253 
254 	/*
255 	 * The following can be overridden after boot via sysctl.  Note:
256 	 * unless overriden, these macros are ultimately based on maxusers.
257 	 * Limit maxproc so that kmap entries cannot be exhausted by
258 	 * processes.
259 	 */
260 	maxproc = NPROC;
261 	TUNABLE_INT_FETCH("kern.maxproc", &maxproc);
262 	if (maxproc > (physpages / 12))
263 		maxproc = physpages / 12;
264 	if (maxproc > pid_max)
265 		maxproc = pid_max;
266 	maxprocperuid = (maxproc * 9) / 10;
267 
268 	/*
269 	 * The default limit for maxfiles is 1/12 of the number of
270 	 * physical page but not less than 16 times maxusers.
271 	 * At most it can be 1/6 the number of physical pages.
272 	 */
273 	maxfiles = imax(MAXFILES, physpages / 8);
274 	TUNABLE_INT_FETCH("kern.maxfiles", &maxfiles);
275 	if (maxfiles > (physpages / 4))
276 		maxfiles = physpages / 4;
277 	maxfilesperproc = (maxfiles / 10) * 9;
278 	TUNABLE_INT_FETCH("kern.maxfilesperproc", &maxfilesperproc);
279 
280 	/*
281 	 * Cannot be changed after boot.
282 	 */
283 	nbuf = NBUF;
284 	TUNABLE_INT_FETCH("kern.nbuf", &nbuf);
285 	TUNABLE_INT_FETCH("kern.bio_transient_maxcnt", &bio_transient_maxcnt);
286 
287 	/*
288 	 * The default for maxpipekva is min(1/64 of the kernel address space,
289 	 * max(1/64 of main memory, 512KB)).  See sys_pipe.c for more details.
290 	 */
291 	maxpipekva = (physpages / 64) * PAGE_SIZE;
292 	TUNABLE_LONG_FETCH("kern.ipc.maxpipekva", &maxpipekva);
293 	if (maxpipekva < 512 * 1024)
294 		maxpipekva = 512 * 1024;
295 	if (maxpipekva > (VM_MAX_KERNEL_ADDRESS - VM_MIN_KERNEL_ADDRESS) / 64)
296 		maxpipekva = (VM_MAX_KERNEL_ADDRESS - VM_MIN_KERNEL_ADDRESS) /
297 		    64;
298 }
299 
300 /*
301  * Sysctl stringifying handler for kern.vm_guest.
302  */
303 static int
304 sysctl_kern_vm_guest(SYSCTL_HANDLER_ARGS)
305 {
306 	return (SYSCTL_OUT_STR(req, vm_guest_sysctl_names[vm_guest]));
307 }
308