sys/kern/subr_param.c

   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  * $FreeBSD: src/sys/kern/subr_param.c,v 1.42.2.10 2002/03/09 21:05:47 silby Exp $
  36  */
  37
  38 #include "opt_param.h"
  39 #include "opt_maxusers.h"
  40
  41 #include <sys/param.h>
  42 #include <sys/systm.h>
  43 #include <sys/kernel.h>
  44 #include <sys/malloc.h>
  45 #include <sys/sysctl.h>
  46 #include <vm/pmap.h>
  47 #include <machine/vmparam.h>
  48
  49 /*
  50  * System parameter formulae.
  51  */
  52
  53 #ifndef HZ_DEFAULT
  54 #define HZ_DEFAULT      100
  55 #endif
  56 #define NPROC           (20 + 16 * maxusers)
  57 #ifndef NBUF
  58 #define NBUF            0
  59 #endif
  60 #ifndef MAXFILES
  61 #define MAXFILES        (maxproc * 16)
  62 #endif
  63 #ifndef MAXPOSIXLOCKSPERUID
  64 #define MAXPOSIXLOCKSPERUID (maxproc * 4)
  65 #endif
  66
  67 static int sysctl_kern_vmm_guest(SYSCTL_HANDLER_ARGS);
  68
  69 int     hz;
  70 int     stathz;
  71 int     profhz;
  72 int     ustick;                         /* tick interval in microseconds */
  73 int     nstick;                         /* tick interval in nanoseconds */
  74 int     maxusers;                       /* base tunable */
  75 int     maxproc;                        /* maximum # of processes */
  76 int     maxprocperuid;                  /* max # of procs per user */
  77 int     maxfiles;                       /* system wide open files limit */
  78 int     maxfilesrootres;                /* descriptors reserved for root use */
  79 int     minfilesperproc;                /* per-proc min open files (safety) */
  80 int     maxfilesperproc;                /* per-proc open files limit */
  81 int     maxfilesperuser;                /* per-user open files limit */
  82 int     maxposixlocksperuid;            /* max # POSIX locks per uid */
  83 int     ncallout;                       /* maximum # of timer events */
  84 int     mbuf_wait = 32;                 /* mbuf sleep time in ticks */
  85 long    nbuf;
  86 long    nswbuf_mem;
  87 long    nswbuf_kva;
  88 long    nswbuf_raw;
  89 long    maxswzone;                      /* max swmeta KVA storage */
  90 long    maxbcache;                      /* max buffer cache KVA storage */
  91 enum vmm_guest_type vmm_guest = VMM_GUEST_NONE; /* Running as VM guest? */
  92 u_quad_t        maxtsiz;                        /* max text size */
  93 u_quad_t        dfldsiz;                        /* initial data size limit */
  94 u_quad_t        maxdsiz;                        /* max data size */
  95 u_quad_t        dflssiz;                        /* initial stack size limit */
  96 u_quad_t        maxssiz;                        /* max stack size */
  97 u_quad_t        sgrowsiz;                       /* amount to grow stack */
  98
  99 SYSCTL_PROC(_kern, OID_AUTO, vmm_guest, CTLFLAG_RD | CTLTYPE_STRING,
 100     NULL, 0, sysctl_kern_vmm_guest, "A",
 101     "Virtual machine guest type");
 102
 103 /*
 104  * These have to be allocated somewhere; allocating
 105  * them here forces loader errors if this file is omitted
 106  * (if they've been externed everywhere else; hah!).
 107  */
 108 struct  buf *swbuf_mem;
 109 struct  buf *swbuf_kva;
 110 struct  buf *swbuf_raw;
 111
 112 struct vmm_bname {
 113         const char *str;
 114         enum vmm_guest_type type;
 115 };
 116
 117 static struct vmm_bname vmm_bnames[] = {
 118         { "QEMU",       VMM_GUEST_QEMU },       /* QEMU */
 119         { "Plex86",     VMM_GUEST_PLEX86 },     /* Plex86 */
 120         { "Bochs",      VMM_GUEST_BOCHS },      /* Bochs */
 121         { "Xen",        VMM_GUEST_XEN },        /* Xen */
 122         { "BHYVE",      VMM_GUEST_BHYVE },      /* bhyve */
 123         { "Seabios",    VMM_GUEST_KVM},         /* KVM */
 124         { NULL, 0 }
 125 };
 126
 127 static struct vmm_bname vmm_pnames[] = {
 128         { "VMware Virtual Platform",    VMM_GUEST_VMWARE },     /* VMWare VM */
 129         { "Virtual Machine",            VMM_GUEST_HYPERV },     /* MS Hyper-V */
 130         { "VirtualBox",                 VMM_GUEST_VBOX },       /* Sun VirtualBox */
 131         { "Parallels Virtual Platform", VMM_GUEST_PARALLELS },  /* Parallels VM */
 132         { "KVM",                        VMM_GUEST_KVM },        /* KVM */
 133         { NULL, 0 }
 134 };
 135
 136 static const char *const vmm_guest_sysctl_names[] = {
 137         "none",
 138         "qemu",
 139         "plex86",
 140         "bochs",
 141         "xen",
 142         "bhyve",
 143         "kvm",
 144         "vmware",
 145         "hyperv",
 146         "vbox",
 147         "parallels",
 148         "vkernel",
 149         "unknown",
 150         NULL
 151 };
 152 CTASSERT(NELEM(vmm_guest_sysctl_names) - 1 == VMM_GUEST_LAST);
 153
 154 char            vmm_vendor[16];
 155 SYSCTL_STRING(_kern, OID_AUTO, vmm_vendor, CTLFLAG_RD, vmm_vendor, 0,
 156     "Virtual machine vendor");
 157
 158 /*
 159  * Detect known Virtual Machine hosts by inspecting the emulated BIOS.
 160  */
 161 enum vmm_guest_type
 162 detect_virtual(void)
 163 {
 164         char *sysenv;
 165         int i;
 166
 167         sysenv = kgetenv("smbios.bios.vendor");
 168         if (sysenv != NULL) {
 169                 for (i = 0; vmm_bnames[i].str != NULL; i++)
 170                         if (strcmp(sysenv, vmm_bnames[i].str) == 0) {
 171                                 kfreeenv(sysenv);
 172                                 return (vmm_bnames[i].type);
 173                         }
 174                 kfreeenv(sysenv);
 175         }
 176         sysenv = kgetenv("smbios.system.product");
 177         if (sysenv != NULL) {
 178                 for (i = 0; vmm_pnames[i].str != NULL; i++)
 179                         if (strcmp(sysenv, vmm_pnames[i].str) == 0) {
 180                                 kfreeenv(sysenv);
 181                                 return (vmm_pnames[i].type);
 182                         }
 183                 kfreeenv(sysenv);
 184         }
 185         return (VMM_GUEST_NONE);
 186 }
 187
 188 /*
 189  * Boot time overrides that are not scaled against main memory
 190  */
 191 void
 192 init_param1(void)
 193 {
 194         hz = HZ_DEFAULT;
 195         TUNABLE_INT_FETCH("kern.hz", &hz);
 196         stathz = hz * 128 / 100;
 197         profhz = stathz;
 198         ustick = 1000000 / hz;
 199         nstick = 1000000000 / hz;
 200         /* can adjust 30ms in 60s */
 201         ntp_default_tick_delta = howmany(30000000, 60 * hz);
 202
 203 #ifdef VM_SWZONE_SIZE_MAX
 204         maxswzone = VM_SWZONE_SIZE_MAX;
 205 #endif
 206         TUNABLE_LONG_FETCH("kern.maxswzone", &maxswzone);
 207 #ifdef VM_BCACHE_SIZE_MAX
 208         maxbcache = VM_BCACHE_SIZE_MAX;
 209 #endif
 210         TUNABLE_LONG_FETCH("kern.maxbcache", &maxbcache);
 211         maxtsiz = MAXTSIZ;
 212         TUNABLE_QUAD_FETCH("kern.maxtsiz", &maxtsiz);
 213         dfldsiz = DFLDSIZ;
 214         TUNABLE_QUAD_FETCH("kern.dfldsiz", &dfldsiz);
 215         maxdsiz = MAXDSIZ;
 216         TUNABLE_QUAD_FETCH("kern.maxdsiz", &maxdsiz);
 217         dflssiz = DFLSSIZ;
 218         TUNABLE_QUAD_FETCH("kern.dflssiz", &dflssiz);
 219         maxssiz = MAXSSIZ;
 220         TUNABLE_QUAD_FETCH("kern.maxssiz", &maxssiz);
 221         sgrowsiz = SGROWSIZ;
 222         TUNABLE_QUAD_FETCH("kern.sgrowsiz", &sgrowsiz);
 223 }
 224
 225 /*
 226  * Boot time overrides that are scaled against main memory
 227  */
 228 void
 229 init_param2(int physpages)
 230 {
 231         size_t limsize;
 232
 233         /*
 234          * Calculate manually becaus the VM page queues / system is not set
 235          * up yet.
 236          */
 237         limsize = (size_t)physpages * PAGE_SIZE;
 238         if (limsize > KvaSize)
 239                 limsize = KvaSize;
 240         if (maxswzone > limsize / 2)    /* maxswzone size (1/2 of phys mem) */
 241                 maxswzone = limsize / 2;
 242
 243         limsize /= 1024 * 1024;         /* smaller of KVM or physmem in MB */
 244
 245         /* Base parameters */
 246         maxusers = MAXUSERS;
 247         TUNABLE_INT_FETCH("kern.maxusers", &maxusers);
 248         if (maxusers == 0) {
 249                 maxusers = limsize / 8;         /* ~384 per 3G */
 250                 if (maxusers < 32)
 251                         maxusers = 32;
 252                 /* no upper limit */
 253         }
 254
 255         /*
 256          * The following can be overridden after boot via sysctl.  Note:
 257          * unless overriden, these macros are ultimately based on maxusers.
 258          *
 259          * Limit maxproc so that kmap entries cannot be exhausted by
 260          * processes.  This limitation can be a bit problematic because
 261          * processes can have a wide range of complexity.
 262          */
 263         maxproc = NPROC;
 264         TUNABLE_INT_FETCH("kern.maxproc", &maxproc);
 265         if (maxproc < 32)
 266                 maxproc = 32;
 267         if (maxproc > limsize * 40)
 268                 maxproc = limsize * 40;
 269
 270         /*
 271          * Maximum number of open files
 272          */
 273         maxfiles = MAXFILES;
 274         TUNABLE_INT_FETCH("kern.maxfiles", &maxfiles);
 275         if (maxfiles < 128)
 276                 maxfiles = 128;
 277
 278         /*
 279          * Limit file descriptors so no single user can exhaust the
 280          * system.
 281          *
 282          * WARNING: Do not set minfilesperproc too high or the user
 283          *          can exhaust the system with a combination of fork()
 284          *          and open().  Actual worst case is:
 285          *
 286          *          (minfilesperproc * maxprocperuid) + maxfilesperuser
 287          */
 288         maxprocperuid = maxproc / 4;
 289         if (maxprocperuid < 128)
 290                 maxprocperuid = maxproc / 2;
 291         minfilesperproc = 8;
 292         maxfilesperproc = maxfiles / 4;
 293         maxfilesperuser = maxfilesperproc * 2;
 294         maxfilesrootres = maxfiles / 20;
 295
 296         /*
 297          * Severe hack to try to prevent pipe() descriptors from
 298          * blowing away kernel memory.
 299          */
 300         if (KvaSize <= (vm_offset_t)(1536LL * 1024 * 1024) &&
 301             maxfilesperuser > 20000) {
 302                 maxfilesperuser = 20000;
 303         }
 304
 305         maxposixlocksperuid = MAXPOSIXLOCKSPERUID;
 306         TUNABLE_INT_FETCH("kern.maxposixlocksperuid", &maxposixlocksperuid);
 307
 308         /*
 309          * Unless overriden, NBUF is typically 0 (auto-sized later).
 310          */
 311         nbuf = NBUF;
 312         TUNABLE_LONG_FETCH("kern.nbuf", &nbuf);
 313
 314         /*
 315          * Calculate the size of the callout wheel.  Limit to approximately
 316          * 5 minutes worth of table (maxproc would have to be pretty huge),
 317          * as more is not likely to gain us anything.
 318          */
 319         ncallout = 16 + maxproc + maxfiles;
 320         if (ncallout > 5*60*hz)
 321                 ncallout = 5*60*hz;
 322         TUNABLE_INT_FETCH("kern.ncallout", &ncallout);
 323 }
 324
 325 /*
 326  * Sysctl stringifying handler for kern.vmm_guest.
 327  */
 328 static int
 329 sysctl_kern_vmm_guest(SYSCTL_HANDLER_ARGS)
 330 {
 331         return (SYSCTL_OUT(req, vmm_guest_sysctl_names[vmm_guest],
 332             strlen(vmm_guest_sysctl_names[vmm_guest])));
 333 }