linux/Documentation/sysctl/kernel.txt
<<
>>
Prefs
   1Documentation for /proc/sys/kernel/*    kernel version 2.2.10
   2        (c) 1998, 1999,  Rik van Riel <riel@nl.linux.org>
   3        (c) 2009,        Shen Feng<shen@cn.fujitsu.com>
   4
   5For general info and legal blurb, please look in README.
   6
   7==============================================================
   8
   9This file contains documentation for the sysctl files in
  10/proc/sys/kernel/ and is valid for Linux kernel version 2.2.
  11
  12The files in this directory can be used to tune and monitor
  13miscellaneous and general things in the operation of the Linux
  14kernel. Since some of the files _can_ be used to screw up your
  15system, it is advisable to read both documentation and source
  16before actually making adjustments.
  17
  18Currently, these files might (depending on your configuration)
  19show up in /proc/sys/kernel:
  20
  21- acct
  22- acpi_video_flags
  23- auto_msgmni
  24- bootloader_type            [ X86 only ]
  25- bootloader_version         [ X86 only ]
  26- callhome                   [ S390 only ]
  27- cap_last_cap
  28- core_pattern
  29- core_pipe_limit
  30- core_uses_pid
  31- ctrl-alt-del
  32- dmesg_restrict
  33- domainname
  34- hostname
  35- hotplug
  36- kptr_restrict
  37- kstack_depth_to_print       [ X86 only ]
  38- l2cr                        [ PPC only ]
  39- modprobe                    ==> Documentation/debugging-modules.txt
  40- modules_disabled
  41- msgmax
  42- msgmnb
  43- msgmni
  44- nmi_watchdog
  45- osrelease
  46- ostype
  47- overflowgid
  48- overflowuid
  49- panic
  50- panic_on_oops
  51- panic_on_unrecovered_nmi
  52- panic_on_stackoverflow
  53- pid_max
  54- powersave-nap               [ PPC only ]
  55- printk
  56- printk_delay
  57- printk_ratelimit
  58- printk_ratelimit_burst
  59- randomize_va_space
  60- real-root-dev               ==> Documentation/initrd.txt
  61- reboot-cmd                  [ SPARC only ]
  62- rtsig-max
  63- rtsig-nr
  64- sem
  65- sg-big-buff                 [ generic SCSI device (sg) ]
  66- shm_rmid_forced
  67- shmall
  68- shmmax                      [ sysv ipc ]
  69- shmmni
  70- softlockup_thresh
  71- stop-a                      [ SPARC only ]
  72- sysrq                       ==> Documentation/sysrq.txt
  73- tainted
  74- threads-max
  75- unknown_nmi_panic
  76- version
  77
  78==============================================================
  79
  80acct:
  81
  82highwater lowwater frequency
  83
  84If BSD-style process accounting is enabled these values control
  85its behaviour. If free space on filesystem where the log lives
  86goes below <lowwater>% accounting suspends. If free space gets
  87above <highwater>% accounting resumes. <Frequency> determines
  88how often do we check the amount of free space (value is in
  89seconds). Default:
  904 2 30
  91That is, suspend accounting if there left <= 2% free; resume it
  92if we got >=4%; consider information about amount of free space
  93valid for 30 seconds.
  94
  95==============================================================
  96
  97acpi_video_flags:
  98
  99flags
 100
 101See Doc*/kernel/power/video.txt, it allows mode of video boot to be
 102set during run time.
 103
 104==============================================================
 105
 106auto_msgmni:
 107
 108Enables/Disables automatic recomputing of msgmni upon memory add/remove
 109or upon ipc namespace creation/removal (see the msgmni description
 110above). Echoing "1" into this file enables msgmni automatic recomputing.
 111Echoing "0" turns it off. auto_msgmni default value is 1.
 112
 113
 114==============================================================
 115
 116bootloader_type:
 117
 118x86 bootloader identification
 119
 120This gives the bootloader type number as indicated by the bootloader,
 121shifted left by 4, and OR'd with the low four bits of the bootloader
 122version.  The reason for this encoding is that this used to match the
 123type_of_loader field in the kernel header; the encoding is kept for
 124backwards compatibility.  That is, if the full bootloader type number
 125is 0x15 and the full version number is 0x234, this file will contain
 126the value 340 = 0x154.
 127
 128See the type_of_loader and ext_loader_type fields in
 129Documentation/x86/boot.txt for additional information.
 130
 131==============================================================
 132
 133bootloader_version:
 134
 135x86 bootloader version
 136
 137The complete bootloader version number.  In the example above, this
 138file will contain the value 564 = 0x234.
 139
 140See the type_of_loader and ext_loader_ver fields in
 141Documentation/x86/boot.txt for additional information.
 142
 143==============================================================
 144
 145callhome:
 146
 147Controls the kernel's callhome behavior in case of a kernel panic.
 148
 149The s390 hardware allows an operating system to send a notification
 150to a service organization (callhome) in case of an operating system panic.
 151
 152When the value in this file is 0 (which is the default behavior)
 153nothing happens in case of a kernel panic. If this value is set to "1"
 154the complete kernel oops message is send to the IBM customer service
 155organization in case the mainframe the Linux operating system is running
 156on has a service contract with IBM.
 157
 158==============================================================
 159
 160cap_last_cap
 161
 162Highest valid capability of the running kernel.  Exports
 163CAP_LAST_CAP from the kernel.
 164
 165==============================================================
 166
 167core_pattern:
 168
 169core_pattern is used to specify a core dumpfile pattern name.
 170. max length 128 characters; default value is "core"
 171. core_pattern is used as a pattern template for the output filename;
 172  certain string patterns (beginning with '%') are substituted with
 173  their actual values.
 174. backward compatibility with core_uses_pid:
 175        If core_pattern does not include "%p" (default does not)
 176        and core_uses_pid is set, then .PID will be appended to
 177        the filename.
 178. corename format specifiers:
 179        %<NUL>  '%' is dropped
 180        %%      output one '%'
 181        %p      pid
 182        %u      uid
 183        %g      gid
 184        %d      dump mode, matches PR_SET_DUMPABLE and
 185                /proc/sys/fs/suid_dumpable
 186        %s      signal number
 187        %t      UNIX time of dump
 188        %h      hostname
 189        %e      executable filename (may be shortened)
 190        %E      executable path
 191        %<OTHER> both are dropped
 192. If the first character of the pattern is a '|', the kernel will treat
 193  the rest of the pattern as a command to run.  The core dump will be
 194  written to the standard input of that program instead of to a file.
 195
 196==============================================================
 197
 198core_pipe_limit:
 199
 200This sysctl is only applicable when core_pattern is configured to pipe
 201core files to a user space helper (when the first character of
 202core_pattern is a '|', see above).  When collecting cores via a pipe
 203to an application, it is occasionally useful for the collecting
 204application to gather data about the crashing process from its
 205/proc/pid directory.  In order to do this safely, the kernel must wait
 206for the collecting process to exit, so as not to remove the crashing
 207processes proc files prematurely.  This in turn creates the
 208possibility that a misbehaving userspace collecting process can block
 209the reaping of a crashed process simply by never exiting.  This sysctl
 210defends against that.  It defines how many concurrent crashing
 211processes may be piped to user space applications in parallel.  If
 212this value is exceeded, then those crashing processes above that value
 213are noted via the kernel log and their cores are skipped.  0 is a
 214special value, indicating that unlimited processes may be captured in
 215parallel, but that no waiting will take place (i.e. the collecting
 216process is not guaranteed access to /proc/<crashing pid>/).  This
 217value defaults to 0.
 218
 219==============================================================
 220
 221core_uses_pid:
 222
 223The default coredump filename is "core".  By setting
 224core_uses_pid to 1, the coredump filename becomes core.PID.
 225If core_pattern does not include "%p" (default does not)
 226and core_uses_pid is set, then .PID will be appended to
 227the filename.
 228
 229==============================================================
 230
 231ctrl-alt-del:
 232
 233When the value in this file is 0, ctrl-alt-del is trapped and
 234sent to the init(1) program to handle a graceful restart.
 235When, however, the value is > 0, Linux's reaction to a Vulcan
 236Nerve Pinch (tm) will be an immediate reboot, without even
 237syncing its dirty buffers.
 238
 239Note: when a program (like dosemu) has the keyboard in 'raw'
 240mode, the ctrl-alt-del is intercepted by the program before it
 241ever reaches the kernel tty layer, and it's up to the program
 242to decide what to do with it.
 243
 244==============================================================
 245
 246dmesg_restrict:
 247
 248This toggle indicates whether unprivileged users are prevented
 249from using dmesg(8) to view messages from the kernel's log buffer.
 250When dmesg_restrict is set to (0) there are no restrictions. When
 251dmesg_restrict is set set to (1), users must have CAP_SYSLOG to use
 252dmesg(8).
 253
 254The kernel config option CONFIG_SECURITY_DMESG_RESTRICT sets the
 255default value of dmesg_restrict.
 256
 257==============================================================
 258
 259domainname & hostname:
 260
 261These files can be used to set the NIS/YP domainname and the
 262hostname of your box in exactly the same way as the commands
 263domainname and hostname, i.e.:
 264# echo "darkstar" > /proc/sys/kernel/hostname
 265# echo "mydomain" > /proc/sys/kernel/domainname
 266has the same effect as
 267# hostname "darkstar"
 268# domainname "mydomain"
 269
 270Note, however, that the classic darkstar.frop.org has the
 271hostname "darkstar" and DNS (Internet Domain Name Server)
 272domainname "frop.org", not to be confused with the NIS (Network
 273Information Service) or YP (Yellow Pages) domainname. These two
 274domain names are in general different. For a detailed discussion
 275see the hostname(1) man page.
 276
 277==============================================================
 278
 279hotplug:
 280
 281Path for the hotplug policy agent.
 282Default value is "/sbin/hotplug".
 283
 284==============================================================
 285
 286kptr_restrict:
 287
 288This toggle indicates whether restrictions are placed on
 289exposing kernel addresses via /proc and other interfaces.  When
 290kptr_restrict is set to (0), there are no restrictions.  When
 291kptr_restrict is set to (1), the default, kernel pointers
 292printed using the %pK format specifier will be replaced with 0's
 293unless the user has CAP_SYSLOG.  When kptr_restrict is set to
 294(2), kernel pointers printed using %pK will be replaced with 0's
 295regardless of privileges.
 296
 297==============================================================
 298
 299kstack_depth_to_print: (X86 only)
 300
 301Controls the number of words to print when dumping the raw
 302kernel stack.
 303
 304==============================================================
 305
 306l2cr: (PPC only)
 307
 308This flag controls the L2 cache of G3 processor boards. If
 3090, the cache is disabled. Enabled if nonzero.
 310
 311==============================================================
 312
 313modules_disabled:
 314
 315A toggle value indicating if modules are allowed to be loaded
 316in an otherwise modular kernel.  This toggle defaults to off
 317(0), but can be set true (1).  Once true, modules can be
 318neither loaded nor unloaded, and the toggle cannot be set back
 319to false.
 320
 321==============================================================
 322
 323nmi_watchdog:
 324
 325Enables/Disables the NMI watchdog on x86 systems. When the value is
 326non-zero the NMI watchdog is enabled and will continuously test all
 327online cpus to determine whether or not they are still functioning
 328properly. Currently, passing "nmi_watchdog=" parameter at boot time is
 329required for this function to work.
 330
 331If LAPIC NMI watchdog method is in use (nmi_watchdog=2 kernel
 332parameter), the NMI watchdog shares registers with oprofile. By
 333disabling the NMI watchdog, oprofile may have more registers to
 334utilize.
 335
 336==============================================================
 337
 338osrelease, ostype & version:
 339
 340# cat osrelease
 3412.1.88
 342# cat ostype
 343Linux
 344# cat version
 345#5 Wed Feb 25 21:49:24 MET 1998
 346
 347The files osrelease and ostype should be clear enough. Version
 348needs a little more clarification however. The '#5' means that
 349this is the fifth kernel built from this source base and the
 350date behind it indicates the time the kernel was built.
 351The only way to tune these values is to rebuild the kernel :-)
 352
 353==============================================================
 354
 355overflowgid & overflowuid:
 356
 357if your architecture did not always support 32-bit UIDs (i.e. arm,
 358i386, m68k, sh, and sparc32), a fixed UID and GID will be returned to
 359applications that use the old 16-bit UID/GID system calls, if the
 360actual UID or GID would exceed 65535.
 361
 362These sysctls allow you to change the value of the fixed UID and GID.
 363The default is 65534.
 364
 365==============================================================
 366
 367panic:
 368
 369The value in this file represents the number of seconds the kernel
 370waits before rebooting on a panic. When you use the software watchdog,
 371the recommended setting is 60.
 372
 373==============================================================
 374
 375panic_on_unrecovered_nmi:
 376
 377The default Linux behaviour on an NMI of either memory or unknown is
 378to continue operation. For many environments such as scientific
 379computing it is preferable that the box is taken out and the error
 380dealt with than an uncorrected parity/ECC error get propagated.
 381
 382A small number of systems do generate NMI's for bizarre random reasons
 383such as power management so the default is off. That sysctl works like
 384the existing panic controls already in that directory.
 385
 386==============================================================
 387
 388panic_on_oops:
 389
 390Controls the kernel's behaviour when an oops or BUG is encountered.
 391
 3920: try to continue operation
 393
 3941: panic immediately.  If the `panic' sysctl is also non-zero then the
 395   machine will be rebooted.
 396
 397==============================================================
 398
 399panic_on_stackoverflow:
 400
 401Controls the kernel's behavior when detecting the overflows of
 402kernel, IRQ and exception stacks except a user stack.
 403This file shows up if CONFIG_DEBUG_STACKOVERFLOW is enabled.
 404
 4050: try to continue operation.
 406
 4071: panic immediately.
 408
 409==============================================================
 410
 411
 412pid_max:
 413
 414PID allocation wrap value.  When the kernel's next PID value
 415reaches this value, it wraps back to a minimum PID value.
 416PIDs of value pid_max or larger are not allocated.
 417
 418==============================================================
 419
 420ns_last_pid:
 421
 422The last pid allocated in the current (the one task using this sysctl
 423lives in) pid namespace. When selecting a pid for a next task on fork
 424kernel tries to allocate a number starting from this one.
 425
 426==============================================================
 427
 428powersave-nap: (PPC only)
 429
 430If set, Linux-PPC will use the 'nap' mode of powersaving,
 431otherwise the 'doze' mode will be used.
 432
 433==============================================================
 434
 435printk:
 436
 437The four values in printk denote: console_loglevel,
 438default_message_loglevel, minimum_console_loglevel and
 439default_console_loglevel respectively.
 440
 441These values influence printk() behavior when printing or
 442logging error messages. See 'man 2 syslog' for more info on
 443the different loglevels.
 444
 445- console_loglevel: messages with a higher priority than
 446  this will be printed to the console
 447- default_message_loglevel: messages without an explicit priority
 448  will be printed with this priority
 449- minimum_console_loglevel: minimum (highest) value to which
 450  console_loglevel can be set
 451- default_console_loglevel: default value for console_loglevel
 452
 453==============================================================
 454
 455printk_delay:
 456
 457Delay each printk message in printk_delay milliseconds
 458
 459Value from 0 - 10000 is allowed.
 460
 461==============================================================
 462
 463printk_ratelimit:
 464
 465Some warning messages are rate limited. printk_ratelimit specifies
 466the minimum length of time between these messages (in jiffies), by
 467default we allow one every 5 seconds.
 468
 469A value of 0 will disable rate limiting.
 470
 471==============================================================
 472
 473printk_ratelimit_burst:
 474
 475While long term we enforce one message per printk_ratelimit
 476seconds, we do allow a burst of messages to pass through.
 477printk_ratelimit_burst specifies the number of messages we can
 478send before ratelimiting kicks in.
 479
 480==============================================================
 481
 482randomize_va_space:
 483
 484This option can be used to select the type of process address
 485space randomization that is used in the system, for architectures
 486that support this feature.
 487
 4880 - Turn the process address space randomization off.  This is the
 489    default for architectures that do not support this feature anyways,
 490    and kernels that are booted with the "norandmaps" parameter.
 491
 4921 - Make the addresses of mmap base, stack and VDSO page randomized.
 493    This, among other things, implies that shared libraries will be
 494    loaded to random addresses.  Also for PIE-linked binaries, the
 495    location of code start is randomized.  This is the default if the
 496    CONFIG_COMPAT_BRK option is enabled.
 497
 4982 - Additionally enable heap randomization.  This is the default if
 499    CONFIG_COMPAT_BRK is disabled.
 500
 501    There are a few legacy applications out there (such as some ancient
 502    versions of libc.so.5 from 1996) that assume that brk area starts
 503    just after the end of the code+bss.  These applications break when
 504    start of the brk area is randomized.  There are however no known
 505    non-legacy applications that would be broken this way, so for most
 506    systems it is safe to choose full randomization.
 507
 508    Systems with ancient and/or broken binaries should be configured
 509    with CONFIG_COMPAT_BRK enabled, which excludes the heap from process
 510    address space randomization.
 511
 512==============================================================
 513
 514reboot-cmd: (Sparc only)
 515
 516??? This seems to be a way to give an argument to the Sparc
 517ROM/Flash boot loader. Maybe to tell it what to do after
 518rebooting. ???
 519
 520==============================================================
 521
 522rtsig-max & rtsig-nr:
 523
 524The file rtsig-max can be used to tune the maximum number
 525of POSIX realtime (queued) signals that can be outstanding
 526in the system.
 527
 528rtsig-nr shows the number of RT signals currently queued.
 529
 530==============================================================
 531
 532sg-big-buff:
 533
 534This file shows the size of the generic SCSI (sg) buffer.
 535You can't tune it just yet, but you could change it on
 536compile time by editing include/scsi/sg.h and changing
 537the value of SG_BIG_BUFF.
 538
 539There shouldn't be any reason to change this value. If
 540you can come up with one, you probably know what you
 541are doing anyway :)
 542
 543==============================================================
 544
 545shmmax:
 546
 547This value can be used to query and set the run time limit
 548on the maximum shared memory segment size that can be created.
 549Shared memory segments up to 1Gb are now supported in the
 550kernel.  This value defaults to SHMMAX.
 551
 552==============================================================
 553
 554shm_rmid_forced:
 555
 556Linux lets you set resource limits, including how much memory one
 557process can consume, via setrlimit(2).  Unfortunately, shared memory
 558segments are allowed to exist without association with any process, and
 559thus might not be counted against any resource limits.  If enabled,
 560shared memory segments are automatically destroyed when their attach
 561count becomes zero after a detach or a process termination.  It will
 562also destroy segments that were created, but never attached to, on exit
 563from the process.  The only use left for IPC_RMID is to immediately
 564destroy an unattached segment.  Of course, this breaks the way things are
 565defined, so some applications might stop working.  Note that this
 566feature will do you no good unless you also configure your resource
 567limits (in particular, RLIMIT_AS and RLIMIT_NPROC).  Most systems don't
 568need this.
 569
 570Note that if you change this from 0 to 1, already created segments
 571without users and with a dead originative process will be destroyed.
 572
 573==============================================================
 574
 575softlockup_thresh:
 576
 577This value can be used to lower the softlockup tolerance threshold.  The
 578default threshold is 60 seconds.  If a cpu is locked up for 60 seconds,
 579the kernel complains.  Valid values are 1-60 seconds.  Setting this
 580tunable to zero will disable the softlockup detection altogether.
 581
 582==============================================================
 583
 584tainted:
 585
 586Non-zero if the kernel has been tainted.  Numeric values, which
 587can be ORed together:
 588
 589   1 - A module with a non-GPL license has been loaded, this
 590       includes modules with no license.
 591       Set by modutils >= 2.4.9 and module-init-tools.
 592   2 - A module was force loaded by insmod -f.
 593       Set by modutils >= 2.4.9 and module-init-tools.
 594   4 - Unsafe SMP processors: SMP with CPUs not designed for SMP.
 595   8 - A module was forcibly unloaded from the system by rmmod -f.
 596  16 - A hardware machine check error occurred on the system.
 597  32 - A bad page was discovered on the system.
 598  64 - The user has asked that the system be marked "tainted".  This
 599       could be because they are running software that directly modifies
 600       the hardware, or for other reasons.
 601 128 - The system has died.
 602 256 - The ACPI DSDT has been overridden with one supplied by the user
 603        instead of using the one provided by the hardware.
 604 512 - A kernel warning has occurred.
 6051024 - A module from drivers/staging was loaded.
 6062048 - The system is working around a severe firmware bug.
 6074096 - An out-of-tree module has been loaded.
 608
 609==============================================================
 610
 611unknown_nmi_panic:
 612
 613The value in this file affects behavior of handling NMI. When the
 614value is non-zero, unknown NMI is trapped and then panic occurs. At
 615that time, kernel debugging information is displayed on console.
 616
 617NMI switch that most IA32 servers have fires unknown NMI up, for
 618example.  If a system hangs up, try pressing the NMI switch.
 619
lxr.linux.no kindly hosted by Redpill Linpro AS, provider of Linux consulting and operations services since 1995.