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 * 4. 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 (maxproc * 2)
80#endif
81
82static int sysctl_kern_vm_guest(SYSCTL_HANDLER_ARGS);
83
84int	hz;
85int	tick;
86int	maxusers;			/* base tunable */
87int	maxproc;			/* maximum # of processes */
88int	maxprocperuid;			/* max # of procs per user */
89int	maxfiles;			/* sys. wide open files limit */
90int	maxfilesperproc;		/* per-proc open files limit */
91int	msgbufsize;			/* size of kernel message buffer */
92int	ncallout;			/* maximum # of timer events */
93int	nbuf;
94int	bio_transient_maxcnt;
95int	ngroups_max;			/* max # groups per process */
96int	nswbuf;
97pid_t	pid_max = PID_MAX;
98long	maxswzone;			/* max swmeta KVA storage */
99long	maxbcache;			/* max buffer cache KVA storage */
100long	maxpipekva;			/* Limit on pipe KVA */
101int 	vm_guest;			/* Running as virtual machine guest? */
102u_long	maxtsiz;			/* max text size */
103u_long	dfldsiz;			/* initial data size limit */
104u_long	maxdsiz;			/* max data size */
105u_long	dflssiz;			/* initial stack size limit */
106u_long	maxssiz;			/* max stack size */
107u_long	sgrowsiz;			/* amount to grow stack */
108
109SYSCTL_INT(_kern, OID_AUTO, hz, CTLFLAG_RDTUN, &hz, 0,
110    "Number of clock ticks per second");
111SYSCTL_INT(_kern, OID_AUTO, ncallout, CTLFLAG_RDTUN, &ncallout, 0,
112    "Number of pre-allocated timer events");
113SYSCTL_INT(_kern, OID_AUTO, nbuf, CTLFLAG_RDTUN, &nbuf, 0,
114    "Number of buffers in the buffer cache");
115SYSCTL_INT(_kern, OID_AUTO, nswbuf, CTLFLAG_RDTUN, &nswbuf, 0,
116    "Number of swap buffers");
117SYSCTL_INT(_kern, OID_AUTO, msgbufsize, CTLFLAG_RDTUN, &msgbufsize, 0,
118    "Size of the kernel message buffer");
119SYSCTL_LONG(_kern, OID_AUTO, maxswzone, CTLFLAG_RDTUN, &maxswzone, 0,
120    "Maximum memory for swap metadata");
121SYSCTL_LONG(_kern, OID_AUTO, maxbcache, CTLFLAG_RDTUN, &maxbcache, 0,
122    "Maximum value of vfs.maxbufspace");
123SYSCTL_INT(_kern, OID_AUTO, bio_transient_maxcnt, CTLFLAG_RDTUN,
124    &bio_transient_maxcnt, 0,
125    "Maximum number of transient BIOs mappings");
126SYSCTL_ULONG(_kern, OID_AUTO, maxtsiz, CTLFLAG_RW | CTLFLAG_TUN, &maxtsiz, 0,
127    "Maximum text size");
128SYSCTL_ULONG(_kern, OID_AUTO, dfldsiz, CTLFLAG_RW | CTLFLAG_TUN, &dfldsiz, 0,
129    "Initial data size limit");
130SYSCTL_ULONG(_kern, OID_AUTO, maxdsiz, CTLFLAG_RW | CTLFLAG_TUN, &maxdsiz, 0,
131    "Maximum data size");
132SYSCTL_ULONG(_kern, OID_AUTO, dflssiz, CTLFLAG_RW | CTLFLAG_TUN, &dflssiz, 0,
133    "Initial stack size limit");
134SYSCTL_ULONG(_kern, OID_AUTO, maxssiz, CTLFLAG_RW | CTLFLAG_TUN, &maxssiz, 0,
135    "Maximum stack size");
136SYSCTL_ULONG(_kern, OID_AUTO, sgrowsiz, CTLFLAG_RW | CTLFLAG_TUN, &sgrowsiz, 0,
137    "Amount to grow stack on a stack fault");
138SYSCTL_PROC(_kern, OID_AUTO, vm_guest, CTLFLAG_RD | CTLTYPE_STRING,
139    NULL, 0, sysctl_kern_vm_guest, "A",
140    "Virtual machine guest detected? (none|generic|xen)");
141
142/*
143 * These have to be allocated somewhere; allocating
144 * them here forces loader errors if this file is omitted
145 * (if they've been externed everywhere else; hah!).
146 */
147struct	buf *swbuf;
148
149/*
150 * The elements of this array are ordered based upon the values of the
151 * corresponding enum VM_GUEST members.
152 */
153static const char *const vm_guest_sysctl_names[] = {
154	"none",
155	"generic",
156	"xen",
157	NULL
158};
159
160#ifndef XEN
161static const char *const vm_bnames[] = {
162	"QEMU",				/* QEMU */
163	"Plex86",			/* Plex86 */
164	"Bochs",			/* Bochs */
165	"Xen",				/* Xen */
166	"BHYVE",			/* bhyve */
167	"Seabios",			/* KVM */
168	NULL
169};
170
171static const char *const vm_pnames[] = {
172	"VMware Virtual Platform",	/* VMWare VM */
173	"Virtual Machine",		/* Microsoft VirtualPC */
174	"VirtualBox",			/* Sun xVM VirtualBox */
175	"Parallels Virtual Platform",	/* Parallels VM */
176	"KVM",				/* KVM */
177	NULL
178};
179
180
181/*
182 * Detect known Virtual Machine hosts by inspecting the emulated BIOS.
183 */
184static enum VM_GUEST
185detect_virtual(void)
186{
187	char *sysenv;
188	int i;
189
190	sysenv = getenv("smbios.bios.vendor");
191	if (sysenv != NULL) {
192		for (i = 0; vm_bnames[i] != NULL; i++)
193			if (strcmp(sysenv, vm_bnames[i]) == 0) {
194				freeenv(sysenv);
195				return (VM_GUEST_VM);
196			}
197		freeenv(sysenv);
198	}
199	sysenv = getenv("smbios.system.product");
200	if (sysenv != NULL) {
201		for (i = 0; vm_pnames[i] != NULL; i++)
202			if (strcmp(sysenv, vm_pnames[i]) == 0) {
203				freeenv(sysenv);
204				return (VM_GUEST_VM);
205			}
206		freeenv(sysenv);
207	}
208	return (VM_GUEST_NO);
209}
210#endif
211
212/*
213 * Boot time overrides that are not scaled against main memory
214 */
215void
216init_param1(void)
217{
218#ifndef XEN
219	vm_guest = detect_virtual();
220#else
221	vm_guest = VM_GUEST_XEN;
222#endif
223	hz = -1;
224	TUNABLE_INT_FETCH("kern.hz", &hz);
225	if (hz == -1)
226		hz = vm_guest > VM_GUEST_NO ? HZ_VM : HZ;
227	tick = 1000000 / hz;
228
229#ifdef VM_SWZONE_SIZE_MAX
230	maxswzone = VM_SWZONE_SIZE_MAX;
231#endif
232	TUNABLE_LONG_FETCH("kern.maxswzone", &maxswzone);
233#ifdef VM_BCACHE_SIZE_MAX
234	maxbcache = VM_BCACHE_SIZE_MAX;
235#endif
236	TUNABLE_LONG_FETCH("kern.maxbcache", &maxbcache);
237	msgbufsize = MSGBUF_SIZE;
238	TUNABLE_INT_FETCH("kern.msgbufsize", &msgbufsize);
239
240	maxtsiz = MAXTSIZ;
241	TUNABLE_ULONG_FETCH("kern.maxtsiz", &maxtsiz);
242	dfldsiz = DFLDSIZ;
243	TUNABLE_ULONG_FETCH("kern.dfldsiz", &dfldsiz);
244	maxdsiz = MAXDSIZ;
245	TUNABLE_ULONG_FETCH("kern.maxdsiz", &maxdsiz);
246	dflssiz = DFLSSIZ;
247	TUNABLE_ULONG_FETCH("kern.dflssiz", &dflssiz);
248	maxssiz = MAXSSIZ;
249	TUNABLE_ULONG_FETCH("kern.maxssiz", &maxssiz);
250	sgrowsiz = SGROWSIZ;
251	TUNABLE_ULONG_FETCH("kern.sgrowsiz", &sgrowsiz);
252
253	/*
254	 * Let the administrator set {NGROUPS_MAX}, but disallow values
255	 * less than NGROUPS_MAX which would violate POSIX.1-2008 or
256	 * greater than INT_MAX-1 which would result in overflow.
257	 */
258	ngroups_max = NGROUPS_MAX;
259	TUNABLE_INT_FETCH("kern.ngroups", &ngroups_max);
260	if (ngroups_max < NGROUPS_MAX)
261		ngroups_max = NGROUPS_MAX;
262
263	/*
264	 * Only allow to lower the maximal pid.
265	 * Prevent setting up a non-bootable system if pid_max is too low.
266	 */
267	TUNABLE_INT_FETCH("kern.pid_max", &pid_max);
268	if (pid_max > PID_MAX)
269		pid_max = PID_MAX;
270	else if (pid_max < 300)
271		pid_max = 300;
272
273	TUNABLE_INT_FETCH("vfs.unmapped_buf_allowed", &unmapped_buf_allowed);
274}
275
276/*
277 * Boot time overrides that are scaled against main memory
278 */
279void
280init_param2(long physpages)
281{
282
283	/* Base parameters */
284	maxusers = MAXUSERS;
285	TUNABLE_INT_FETCH("kern.maxusers", &maxusers);
286	if (maxusers == 0) {
287		maxusers = physpages / (2 * 1024 * 1024 / PAGE_SIZE);
288		if (maxusers < 32)
289			maxusers = 32;
290#ifdef VM_MAX_AUTOTUNE_MAXUSERS
291                if (maxusers > VM_MAX_AUTOTUNE_MAXUSERS)
292                        maxusers = VM_MAX_AUTOTUNE_MAXUSERS;
293#endif
294                /*
295                 * Scales down the function in which maxusers grows once
296                 * we hit 384.
297                 */
298                if (maxusers > 384)
299                        maxusers = 384 + ((maxusers - 384) / 8);
300        }
301
302	/*
303	 * The following can be overridden after boot via sysctl.  Note:
304	 * unless overriden, these macros are ultimately based on maxusers.
305	 * Limit maxproc so that kmap entries cannot be exhausted by
306	 * processes.
307	 */
308	maxproc = NPROC;
309	TUNABLE_INT_FETCH("kern.maxproc", &maxproc);
310	if (maxproc > (physpages / 12))
311		maxproc = physpages / 12;
312	maxprocperuid = (maxproc * 9) / 10;
313
314	/*
315	 * The default limit for maxfiles is 1/12 of the number of
316	 * physical page but not less than 16 times maxusers.
317	 * At most it can be 1/6 the number of physical pages.
318	 */
319	maxfiles = imax(MAXFILES, physpages / 8);
320	TUNABLE_INT_FETCH("kern.maxfiles", &maxfiles);
321	if (maxfiles > (physpages / 4))
322		maxfiles = physpages / 4;
323	maxfilesperproc = (maxfiles / 10) * 9;
324
325	/*
326	 * Cannot be changed after boot.
327	 */
328	nbuf = NBUF;
329	TUNABLE_INT_FETCH("kern.nbuf", &nbuf);
330	TUNABLE_INT_FETCH("kern.bio_transient_maxcnt", &bio_transient_maxcnt);
331
332	/*
333	 * XXX: Does the callout wheel have to be so big?
334	 *
335	 * Clip callout to result of previous function of maxusers maximum
336	 * 384.  This is still huge, but acceptable.
337	 */
338	ncallout = imin(16 + maxproc + maxfiles, 18508);
339	TUNABLE_INT_FETCH("kern.ncallout", &ncallout);
340
341	/*
342	 * The default for maxpipekva is min(1/64 of the kernel address space,
343	 * max(1/64 of main memory, 512KB)).  See sys_pipe.c for more details.
344	 */
345	maxpipekva = (physpages / 64) * PAGE_SIZE;
346	TUNABLE_LONG_FETCH("kern.ipc.maxpipekva", &maxpipekva);
347	if (maxpipekva < 512 * 1024)
348		maxpipekva = 512 * 1024;
349	if (maxpipekva > (VM_MAX_KERNEL_ADDRESS - VM_MIN_KERNEL_ADDRESS) / 64)
350		maxpipekva = (VM_MAX_KERNEL_ADDRESS - VM_MIN_KERNEL_ADDRESS) /
351		    64;
352}
353
354/*
355 * Sysctl stringifying handler for kern.vm_guest.
356 */
357static int
358sysctl_kern_vm_guest(SYSCTL_HANDLER_ARGS)
359{
360	return (SYSCTL_OUT(req, vm_guest_sysctl_names[vm_guest],
361	    strlen(vm_guest_sysctl_names[vm_guest])));
362}
363