1 /*- 2 * SPDX-License-Identifier: BSD-3-Clause 3 * 4 * Copyright (c) 1980, 1986, 1989, 1993 5 * The Regents of the University of California. All rights reserved. 6 * (c) UNIX System Laboratories, Inc. 7 * All or some portions of this file are derived from material licensed 8 * to the University of California by American Telephone and Telegraph 9 * Co. or Unix System Laboratories, Inc. and are reproduced herein with 10 * the permission of UNIX System Laboratories, Inc. 11 * 12 * Redistribution and use in source and binary forms, with or without 13 * modification, are permitted provided that the following conditions 14 * are met: 15 * 1. Redistributions of source code must retain the above copyright 16 * notice, this list of conditions and the following disclaimer. 17 * 2. Redistributions in binary form must reproduce the above copyright 18 * notice, this list of conditions and the following disclaimer in the 19 * documentation and/or other materials provided with the distribution. 20 * 3. Neither the name of the University nor the names of its contributors 21 * may be used to endorse or promote products derived from this software 22 * without specific prior written permission. 23 * 24 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 25 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 26 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 27 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 28 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 29 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 30 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 31 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 32 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 33 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 34 * SUCH DAMAGE. 35 * 36 * @(#)param.c 8.3 (Berkeley) 8/20/94 37 */ 38 39 #include <sys/cdefs.h> 40 __FBSDID("$FreeBSD$"); 41 42 #include "opt_param.h" 43 #include "opt_msgbuf.h" 44 #include "opt_maxphys.h" 45 #include "opt_maxusers.h" 46 47 #include <sys/param.h> 48 #include <sys/systm.h> 49 #include <sys/buf.h> 50 #include <sys/kernel.h> 51 #include <sys/limits.h> 52 #include <sys/msgbuf.h> 53 #include <sys/sysctl.h> 54 #include <sys/proc.h> 55 #include <sys/vnode.h> 56 57 #include <vm/vm.h> 58 #include <vm/vm_param.h> 59 #include <vm/pmap.h> 60 61 /* 62 * System parameter formulae. 63 */ 64 65 #ifndef HZ 66 # define HZ 1000 67 # ifndef HZ_VM 68 # define HZ_VM 100 69 # endif 70 #else 71 # ifndef HZ_VM 72 # define HZ_VM HZ 73 # endif 74 #endif 75 #define NPROC (20 + 16 * maxusers) 76 #ifndef NBUF 77 #define NBUF 0 78 #endif 79 #ifndef MAXFILES 80 #define MAXFILES (40 + 32 * maxusers) 81 #endif 82 83 static int sysctl_kern_vm_guest(SYSCTL_HANDLER_ARGS); 84 85 int hz; /* system clock's frequency */ 86 int tick; /* usec per tick (1000000 / hz) */ 87 struct bintime tick_bt; /* bintime per tick (1s / hz) */ 88 sbintime_t tick_sbt; 89 int maxusers; /* base tunable */ 90 int maxproc; /* maximum # of processes */ 91 int maxprocperuid; /* max # of procs per user */ 92 int maxfiles; /* sys. wide open files limit */ 93 int maxfilesperproc; /* per-proc open files limit */ 94 int msgbufsize; /* size of kernel message buffer */ 95 int nbuf; /* number of bcache bufs */ 96 int bio_transient_maxcnt; 97 int ngroups_max; /* max # groups per process */ 98 int nswbuf; 99 pid_t pid_max = PID_MAX; 100 u_long maxswzone; /* max swmeta KVA storage */ 101 u_long maxbcache; /* max buffer cache KVA storage */ 102 u_long maxpipekva; /* Limit on pipe KVA */ 103 u_long maxphys; /* max raw I/O transfer size */ 104 int vm_guest = VM_GUEST_NO; /* Running as virtual machine guest? */ 105 u_long maxtsiz; /* max text size */ 106 u_long dfldsiz; /* initial data size limit */ 107 u_long maxdsiz; /* max data size */ 108 u_long dflssiz; /* initial stack size limit */ 109 u_long maxssiz; /* max stack size */ 110 u_long sgrowsiz; /* amount to grow stack */ 111 112 SYSCTL_INT(_kern, OID_AUTO, hz, CTLFLAG_RDTUN | CTLFLAG_NOFETCH, &hz, 0, 113 "Number of clock ticks per second"); 114 SYSCTL_INT(_kern, OID_AUTO, nbuf, CTLFLAG_RDTUN | CTLFLAG_NOFETCH, &nbuf, 0, 115 "Number of buffers in the buffer cache"); 116 SYSCTL_INT(_kern, OID_AUTO, nswbuf, CTLFLAG_RDTUN | CTLFLAG_NOFETCH, &nswbuf, 0, 117 "Number of swap buffers"); 118 SYSCTL_INT(_kern, OID_AUTO, msgbufsize, CTLFLAG_RDTUN | CTLFLAG_NOFETCH, &msgbufsize, 0, 119 "Size of the kernel message buffer"); 120 SYSCTL_LONG(_kern, OID_AUTO, maxswzone, CTLFLAG_RDTUN | CTLFLAG_NOFETCH, &maxswzone, 0, 121 "Maximum memory for swap metadata"); 122 SYSCTL_LONG(_kern, OID_AUTO, maxbcache, CTLFLAG_RDTUN | CTLFLAG_NOFETCH, &maxbcache, 0, 123 "Maximum value of vfs.maxbufspace"); 124 SYSCTL_INT(_kern, OID_AUTO, bio_transient_maxcnt, CTLFLAG_RDTUN | CTLFLAG_NOFETCH, 125 &bio_transient_maxcnt, 0, 126 "Maximum number of transient BIOs mappings"); 127 SYSCTL_ULONG(_kern, OID_AUTO, maxtsiz, CTLFLAG_RWTUN | CTLFLAG_NOFETCH, &maxtsiz, 0, 128 "Maximum text size"); 129 SYSCTL_ULONG(_kern, OID_AUTO, dfldsiz, CTLFLAG_RWTUN | CTLFLAG_NOFETCH, &dfldsiz, 0, 130 "Initial data size limit"); 131 SYSCTL_ULONG(_kern, OID_AUTO, maxdsiz, CTLFLAG_RWTUN | CTLFLAG_NOFETCH, &maxdsiz, 0, 132 "Maximum data size"); 133 SYSCTL_ULONG(_kern, OID_AUTO, dflssiz, CTLFLAG_RWTUN | CTLFLAG_NOFETCH, &dflssiz, 0, 134 "Initial stack size limit"); 135 SYSCTL_ULONG(_kern, OID_AUTO, maxssiz, CTLFLAG_RWTUN | CTLFLAG_NOFETCH, &maxssiz, 0, 136 "Maximum stack size"); 137 SYSCTL_ULONG(_kern, OID_AUTO, sgrowsiz, CTLFLAG_RWTUN | CTLFLAG_NOFETCH, &sgrowsiz, 0, 138 "Amount to grow stack on a stack fault"); 139 SYSCTL_PROC(_kern, OID_AUTO, vm_guest, 140 CTLFLAG_RD | CTLTYPE_STRING | CTLFLAG_MPSAFE, NULL, 0, 141 sysctl_kern_vm_guest, "A", 142 "Virtual machine guest detected?"); 143 144 /* 145 * The elements of this array are ordered based upon the values of the 146 * corresponding enum VM_GUEST members. 147 */ 148 static const char *const vm_guest_sysctl_names[] = { 149 [VM_GUEST_NO] = "none", 150 [VM_GUEST_VM] = "generic", 151 [VM_GUEST_XEN] = "xen", 152 [VM_GUEST_HV] = "hv", 153 [VM_GUEST_VMWARE] = "vmware", 154 [VM_GUEST_KVM] = "kvm", 155 [VM_GUEST_BHYVE] = "bhyve", 156 [VM_GUEST_VBOX] = "vbox", 157 [VM_GUEST_PARALLELS] = "parallels", 158 [VM_LAST] = NULL 159 }; 160 CTASSERT(nitems(vm_guest_sysctl_names) - 1 == VM_LAST); 161 162 /* 163 * Boot time overrides that are not scaled against main memory 164 */ 165 void 166 init_param1(void) 167 { 168 169 #if !defined(__mips__) && !defined(__arm64__) 170 TUNABLE_INT_FETCH("kern.kstack_pages", &kstack_pages); 171 #endif 172 hz = -1; 173 TUNABLE_INT_FETCH("kern.hz", &hz); 174 if (hz == -1) 175 hz = vm_guest > VM_GUEST_NO ? HZ_VM : HZ; 176 tick = 1000000 / hz; 177 tick_sbt = SBT_1S / hz; 178 tick_bt = sbttobt(tick_sbt); 179 180 /* 181 * Arrange for ticks to wrap 10 minutes after boot to help catch 182 * sign problems sooner. 183 */ 184 ticks = INT_MAX - (hz * 10 * 60); 185 186 vn_lock_pair_pause_max = hz / 100; 187 if (vn_lock_pair_pause_max == 0) 188 vn_lock_pair_pause_max = 1; 189 190 #ifdef VM_SWZONE_SIZE_MAX 191 maxswzone = VM_SWZONE_SIZE_MAX; 192 #endif 193 TUNABLE_LONG_FETCH("kern.maxswzone", &maxswzone); 194 #ifdef VM_BCACHE_SIZE_MAX 195 maxbcache = VM_BCACHE_SIZE_MAX; 196 #endif 197 TUNABLE_LONG_FETCH("kern.maxbcache", &maxbcache); 198 msgbufsize = MSGBUF_SIZE; 199 TUNABLE_INT_FETCH("kern.msgbufsize", &msgbufsize); 200 201 maxtsiz = MAXTSIZ; 202 TUNABLE_ULONG_FETCH("kern.maxtsiz", &maxtsiz); 203 dfldsiz = DFLDSIZ; 204 TUNABLE_ULONG_FETCH("kern.dfldsiz", &dfldsiz); 205 maxdsiz = MAXDSIZ; 206 TUNABLE_ULONG_FETCH("kern.maxdsiz", &maxdsiz); 207 dflssiz = DFLSSIZ; 208 TUNABLE_ULONG_FETCH("kern.dflssiz", &dflssiz); 209 maxssiz = MAXSSIZ; 210 TUNABLE_ULONG_FETCH("kern.maxssiz", &maxssiz); 211 sgrowsiz = SGROWSIZ; 212 TUNABLE_ULONG_FETCH("kern.sgrowsiz", &sgrowsiz); 213 214 /* 215 * Let the administrator set {NGROUPS_MAX}, but disallow values 216 * less than NGROUPS_MAX which would violate POSIX.1-2008 or 217 * greater than INT_MAX-1 which would result in overflow. 218 */ 219 ngroups_max = NGROUPS_MAX; 220 TUNABLE_INT_FETCH("kern.ngroups", &ngroups_max); 221 if (ngroups_max < NGROUPS_MAX) 222 ngroups_max = NGROUPS_MAX; 223 224 /* 225 * Only allow to lower the maximal pid. 226 * Prevent setting up a non-bootable system if pid_max is too low. 227 */ 228 TUNABLE_INT_FETCH("kern.pid_max", &pid_max); 229 if (pid_max > PID_MAX) 230 pid_max = PID_MAX; 231 else if (pid_max < 300) 232 pid_max = 300; 233 234 TUNABLE_INT_FETCH("vfs.unmapped_buf_allowed", &unmapped_buf_allowed); 235 } 236 237 /* 238 * Boot time overrides that are scaled against main memory 239 */ 240 void 241 init_param2(long physpages) 242 { 243 244 /* Base parameters */ 245 maxusers = MAXUSERS; 246 TUNABLE_INT_FETCH("kern.maxusers", &maxusers); 247 if (maxusers == 0) { 248 maxusers = physpages / (2 * 1024 * 1024 / PAGE_SIZE); 249 if (maxusers < 32) 250 maxusers = 32; 251 #ifdef VM_MAX_AUTOTUNE_MAXUSERS 252 if (maxusers > VM_MAX_AUTOTUNE_MAXUSERS) 253 maxusers = VM_MAX_AUTOTUNE_MAXUSERS; 254 #endif 255 /* 256 * Scales down the function in which maxusers grows once 257 * we hit 384. 258 */ 259 if (maxusers > 384) 260 maxusers = 384 + ((maxusers - 384) / 8); 261 } 262 263 /* 264 * The following can be overridden after boot via sysctl. Note: 265 * unless overridden, these macros are ultimately based on maxusers. 266 * Limit maxproc so that kmap entries cannot be exhausted by 267 * processes. 268 */ 269 maxproc = NPROC; 270 TUNABLE_INT_FETCH("kern.maxproc", &maxproc); 271 if (maxproc > (physpages / 12)) 272 maxproc = physpages / 12; 273 if (maxproc > pid_max) 274 maxproc = pid_max; 275 maxprocperuid = (maxproc * 9) / 10; 276 277 /* 278 * The default limit for maxfiles is 1/12 of the number of 279 * physical page but not less than 16 times maxusers. 280 * At most it can be 1/6 the number of physical pages. 281 */ 282 maxfiles = imax(MAXFILES, physpages / 8); 283 TUNABLE_INT_FETCH("kern.maxfiles", &maxfiles); 284 if (maxfiles > (physpages / 4)) 285 maxfiles = physpages / 4; 286 maxfilesperproc = (maxfiles / 10) * 9; 287 TUNABLE_INT_FETCH("kern.maxfilesperproc", &maxfilesperproc); 288 289 /* 290 * Cannot be changed after boot. 291 */ 292 nbuf = NBUF; 293 TUNABLE_INT_FETCH("kern.nbuf", &nbuf); 294 TUNABLE_INT_FETCH("kern.bio_transient_maxcnt", &bio_transient_maxcnt); 295 maxphys = MAXPHYS; 296 TUNABLE_ULONG_FETCH("kern.maxphys", &maxphys); 297 if (maxphys == 0) { 298 maxphys = MAXPHYS; 299 } else if (__bitcountl(maxphys) != 1) { /* power of two */ 300 if (flsl(maxphys) == NBBY * sizeof(maxphys)) 301 maxphys = MAXPHYS; 302 else 303 maxphys = 1UL << flsl(maxphys); 304 } 305 if (maxphys < PAGE_SIZE) 306 maxphys = MAXPHYS; 307 308 /* 309 * Physical buffers are pre-allocated buffers (struct buf) that 310 * are used as temporary holders for I/O, such as paging I/O. 311 */ 312 TUNABLE_INT_FETCH("kern.nswbuf", &nswbuf); 313 314 /* 315 * The default for maxpipekva is min(1/64 of the kernel address space, 316 * max(1/64 of main memory, 512KB)). See sys_pipe.c for more details. 317 */ 318 maxpipekva = ptoa(physpages / 64); 319 TUNABLE_LONG_FETCH("kern.ipc.maxpipekva", &maxpipekva); 320 if (maxpipekva < 512 * 1024) 321 maxpipekva = 512 * 1024; 322 if (maxpipekva > (VM_MAX_KERNEL_ADDRESS - VM_MIN_KERNEL_ADDRESS) / 64) 323 maxpipekva = (VM_MAX_KERNEL_ADDRESS - VM_MIN_KERNEL_ADDRESS) / 324 64; 325 } 326 327 /* 328 * Sysctl stringifying handler for kern.vm_guest. 329 */ 330 static int 331 sysctl_kern_vm_guest(SYSCTL_HANDLER_ARGS) 332 { 333 return (SYSCTL_OUT_STR(req, vm_guest_sysctl_names[vm_guest])); 334 } 335