1 ===================================
2 Documentation for /proc/sys/kernel/
3 ===================================
5 .. See scripts/check-sysctl-docs to keep this up to date
8 Copyright (c) 1998, 1999, Rik van Riel <riel@nl.linux.org>
10 Copyright (c) 2009, Shen Feng<shen@cn.fujitsu.com>
12 For general info and legal blurb, please look in :doc:`index`.
14 ------------------------------------------------------------------------------
16 This file contains documentation for the sysctl files in
17 ``/proc/sys/kernel/`` and is valid for Linux kernel version 2.2.
19 The files in this directory can be used to tune and monitor
20 miscellaneous and general things in the operation of the Linux
21 kernel. Since some of the files *can* be used to screw up your
22 system, it is advisable to read both documentation and source
23 before actually making adjustments.
25 Currently, these files might (depending on your configuration)
26 show up in ``/proc/sys/kernel``:
36 highwater lowwater frequency
38 If BSD-style process accounting is enabled these values control
39 its behaviour. If free space on filesystem where the log lives
40 goes below ``lowwater``% accounting suspends. If free space gets
41 above ``highwater``% accounting resumes. ``frequency`` determines
42 how often do we check the amount of free space (value is in
49 That is, suspend accounting if free space drops below 2%; resume it
50 if it increases to at least 4%; consider information about amount of
51 free space valid for 30 seconds.
57 See :doc:`/power/video`. This allows the video resume mode to be set,
58 in a similar fashion to the ``acpi_sleep`` kernel parameter, by
59 combining the following values:
71 This variable has no effect and may be removed in future kernel
72 releases. Reading it always returns 0.
73 Up to Linux 3.17, it enabled/disabled automatic recomputing of
75 upon memory add/remove or upon IPC namespace creation/removal.
76 Echoing "1" into this file enabled msgmni automatic recomputing.
77 Echoing "0" turned it off. The default value was 1.
80 bootloader_type (x86 only)
81 ==========================
83 This gives the bootloader type number as indicated by the bootloader,
84 shifted left by 4, and OR'd with the low four bits of the bootloader
85 version. The reason for this encoding is that this used to match the
86 ``type_of_loader`` field in the kernel header; the encoding is kept for
87 backwards compatibility. That is, if the full bootloader type number
88 is 0x15 and the full version number is 0x234, this file will contain
89 the value 340 = 0x154.
91 See the ``type_of_loader`` and ``ext_loader_type`` fields in
92 :doc:`/x86/boot` for additional information.
95 bootloader_version (x86 only)
96 =============================
98 The complete bootloader version number. In the example above, this
99 file will contain the value 564 = 0x234.
101 See the ``type_of_loader`` and ``ext_loader_ver`` fields in
102 :doc:`/x86/boot` for additional information.
108 Highest valid capability of the running kernel. Exports
109 ``CAP_LAST_CAP`` from the kernel.
115 ``core_pattern`` is used to specify a core dumpfile pattern name.
117 * max length 127 characters; default value is "core"
118 * ``core_pattern`` is used as a pattern template for the output
119 filename; certain string patterns (beginning with '%') are
120 substituted with their actual values.
121 * backward compatibility with ``core_uses_pid``:
123 If ``core_pattern`` does not include "%p" (default does not)
124 and ``core_uses_pid`` is set, then .PID will be appended to
127 * corename format specifiers
129 ======== ==========================================
130 %<NUL> '%' is dropped
133 %P global pid (init PID namespace)
135 %I global tid (init PID namespace)
136 %u uid (in initial user namespace)
137 %g gid (in initial user namespace)
138 %d dump mode, matches ``PR_SET_DUMPABLE`` and
139 ``/proc/sys/fs/suid_dumpable``
143 %e executable filename (may be shortened)
145 %c maximum size of core file by resource limit RLIMIT_CORE
146 %<OTHER> both are dropped
147 ======== ==========================================
149 * If the first character of the pattern is a '|', the kernel will treat
150 the rest of the pattern as a command to run. The core dump will be
151 written to the standard input of that program instead of to a file.
157 This sysctl is only applicable when `core_pattern`_ is configured to
158 pipe core files to a user space helper (when the first character of
159 ``core_pattern`` is a '|', see above).
160 When collecting cores via a pipe to an application, it is occasionally
161 useful for the collecting application to gather data about the
162 crashing process from its ``/proc/pid`` directory.
163 In order to do this safely, the kernel must wait for the collecting
164 process to exit, so as not to remove the crashing processes proc files
166 This in turn creates the possibility that a misbehaving userspace
167 collecting process can block the reaping of a crashed process simply
169 This sysctl defends against that.
170 It defines how many concurrent crashing processes may be piped to user
171 space applications in parallel.
172 If this value is exceeded, then those crashing processes above that
173 value are noted via the kernel log and their cores are skipped.
174 0 is a special value, indicating that unlimited processes may be
175 captured in parallel, but that no waiting will take place (i.e. the
176 collecting process is not guaranteed access to ``/proc/<crashing
178 This value defaults to 0.
184 The default coredump filename is "core". By setting
185 ``core_uses_pid`` to 1, the coredump filename becomes core.PID.
186 If `core_pattern`_ does not include "%p" (default does not)
187 and ``core_uses_pid`` is set, then .PID will be appended to
194 When the value in this file is 0, ctrl-alt-del is trapped and
195 sent to the ``init(1)`` program to handle a graceful restart.
196 When, however, the value is > 0, Linux's reaction to a Vulcan
197 Nerve Pinch (tm) will be an immediate reboot, without even
198 syncing its dirty buffers.
201 when a program (like dosemu) has the keyboard in 'raw'
202 mode, the ctrl-alt-del is intercepted by the program before it
203 ever reaches the kernel tty layer, and it's up to the program
204 to decide what to do with it.
210 This toggle indicates whether unprivileged users are prevented
211 from using ``dmesg(8)`` to view messages from the kernel's log
213 When ``dmesg_restrict`` is set to 0 there are no restrictions.
214 When ``dmesg_restrict`` is set set to 1, users must have
215 ``CAP_SYSLOG`` to use ``dmesg(8)``.
217 The kernel config option ``CONFIG_SECURITY_DMESG_RESTRICT`` sets the
218 default value of ``dmesg_restrict``.
221 domainname & hostname
222 =====================
224 These files can be used to set the NIS/YP domainname and the
225 hostname of your box in exactly the same way as the commands
226 domainname and hostname, i.e.::
228 # echo "darkstar" > /proc/sys/kernel/hostname
229 # echo "mydomain" > /proc/sys/kernel/domainname
231 has the same effect as::
233 # hostname "darkstar"
234 # domainname "mydomain"
236 Note, however, that the classic darkstar.frop.org has the
237 hostname "darkstar" and DNS (Internet Domain Name Server)
238 domainname "frop.org", not to be confused with the NIS (Network
239 Information Service) or YP (Yellow Pages) domainname. These two
240 domain names are in general different. For a detailed discussion
241 see the ``hostname(1)`` man page.
244 hardlockup_all_cpu_backtrace
245 ============================
247 This value controls the hard lockup detector behavior when a hard
248 lockup condition is detected as to whether or not to gather further
249 debug information. If enabled, arch-specific all-CPU stack dumping
252 = ============================================
253 0 Do nothing. This is the default behavior.
254 1 On detection capture more debug information.
255 = ============================================
261 This parameter can be used to control whether the kernel panics
262 when a hard lockup is detected.
264 = ===========================
265 0 Don't panic on hard lockup.
266 1 Panic on hard lockup.
267 = ===========================
269 See :doc:`/admin-guide/lockup-watchdogs` for more information.
270 This can also be set using the nmi_watchdog kernel parameter.
276 Path for the hotplug policy agent.
277 Default value is "``/sbin/hotplug``".
283 Controls the kernel's behavior when a hung task is detected.
284 This file shows up if ``CONFIG_DETECT_HUNG_TASK`` is enabled.
286 = =================================================
287 0 Continue operation. This is the default behavior.
289 = =================================================
292 hung_task_check_count
293 =====================
295 The upper bound on the number of tasks that are checked.
296 This file shows up if ``CONFIG_DETECT_HUNG_TASK`` is enabled.
299 hung_task_timeout_secs
300 ======================
302 When a task in D state did not get scheduled
303 for more than this value report a warning.
304 This file shows up if ``CONFIG_DETECT_HUNG_TASK`` is enabled.
306 0 means infinite timeout, no checking is done.
308 Possible values to set are in range {0:``LONG_MAX``/``HZ``}.
311 hung_task_check_interval_secs
312 =============================
314 Hung task check interval. If hung task checking is enabled
315 (see `hung_task_timeout_secs`_), the check is done every
316 ``hung_task_check_interval_secs`` seconds.
317 This file shows up if ``CONFIG_DETECT_HUNG_TASK`` is enabled.
319 0 (default) means use ``hung_task_timeout_secs`` as checking
322 Possible values to set are in range {0:``LONG_MAX``/``HZ``}.
328 The maximum number of warnings to report. During a check interval
329 if a hung task is detected, this value is decreased by 1.
330 When this value reaches 0, no more warnings will be reported.
331 This file shows up if ``CONFIG_DETECT_HUNG_TASK`` is enabled.
333 -1: report an infinite number of warnings.
336 hyperv_record_panic_msg
337 =======================
339 Controls whether the panic kmsg data should be reported to Hyper-V.
341 = =========================================================
342 0 Do not report panic kmsg data.
343 1 Report the panic kmsg data. This is the default behavior.
344 = =========================================================
350 A toggle indicating if the ``kexec_load`` syscall has been disabled.
351 This value defaults to 0 (false: ``kexec_load`` enabled), but can be
352 set to 1 (true: ``kexec_load`` disabled).
353 Once true, kexec can no longer be used, and the toggle cannot be set
355 This allows a kexec image to be loaded before disabling the syscall,
356 allowing a system to set up (and later use) an image without it being
358 Generally used together with the `modules_disabled`_ sysctl.
364 This toggle indicates whether restrictions are placed on
365 exposing kernel addresses via ``/proc`` and other interfaces.
367 When ``kptr_restrict`` is set to 0 (the default) the address is hashed
369 (This is the equivalent to %p.)
371 When ``kptr_restrict`` is set to 1, kernel pointers printed using the
372 %pK format specifier will be replaced with 0s unless the user has
373 ``CAP_SYSLOG`` and effective user and group ids are equal to the real
375 This is because %pK checks are done at read() time rather than open()
376 time, so if permissions are elevated between the open() and the read()
377 (e.g via a setuid binary) then %pK will not leak kernel pointers to
379 Note, this is a temporary solution only.
380 The correct long-term solution is to do the permission checks at
382 Consider removing world read permissions from files that use %pK, and
383 using `dmesg_restrict`_ to protect against uses of %pK in ``dmesg(8)``
384 if leaking kernel pointer values to unprivileged users is a concern.
386 When ``kptr_restrict`` is set to 2, kernel pointers printed using
387 %pK will be replaced with 0s regardless of privileges.
393 The full path to the usermode helper for autoloading kernel modules,
394 by default "/sbin/modprobe". This binary is executed when the kernel
395 requests a module. For example, if userspace passes an unknown
396 filesystem type to mount(), then the kernel will automatically request
397 the corresponding filesystem module by executing this usermode helper.
398 This usermode helper should insert the needed module into the kernel.
400 This sysctl only affects module autoloading. It has no effect on the
401 ability to explicitly insert modules.
403 This sysctl can be used to debug module loading requests::
405 echo '#! /bin/sh' > /tmp/modprobe
406 echo 'echo "$@" >> /tmp/modprobe.log' >> /tmp/modprobe
407 echo 'exec /sbin/modprobe "$@"' >> /tmp/modprobe
408 chmod a+x /tmp/modprobe
409 echo /tmp/modprobe > /proc/sys/kernel/modprobe
411 Alternatively, if this sysctl is set to the empty string, then module
412 autoloading is completely disabled. The kernel will not try to
413 execute a usermode helper at all, nor will it call the
414 kernel_module_request LSM hook.
416 If CONFIG_STATIC_USERMODEHELPER=y is set in the kernel configuration,
417 then the configured static usermode helper overrides this sysctl,
418 except that the empty string is still accepted to completely disable
419 module autoloading as described above.
424 A toggle value indicating if modules are allowed to be loaded
425 in an otherwise modular kernel. This toggle defaults to off
426 (0), but can be set true (1). Once true, modules can be
427 neither loaded nor unloaded, and the toggle cannot be set back
428 to false. Generally used with the `kexec_load_disabled`_ toggle.
433 msgmax, msgmnb, and msgmni
434 ==========================
436 ``msgmax`` is the maximum size of an IPC message, in bytes. 8192 by
437 default (``MSGMAX``).
439 ``msgmnb`` is the maximum size of an IPC queue, in bytes. 16384 by
440 default (``MSGMNB``).
442 ``msgmni`` is the maximum number of IPC queues. 32000 by default
446 msg_next_id, sem_next_id, and shm_next_id (System V IPC)
447 ========================================================
449 These three toggles allows to specify desired id for next allocated IPC
450 object: message, semaphore or shared memory respectively.
452 By default they are equal to -1, which means generic allocation logic.
453 Possible values to set are in range {0:``INT_MAX``}.
456 1) kernel doesn't guarantee, that new object will have desired id. So,
457 it's up to userspace, how to handle an object with "wrong" id.
458 2) Toggle with non-default value will be set back to -1 by kernel after
459 successful IPC object allocation. If an IPC object allocation syscall
460 fails, it is undefined if the value remains unmodified or is reset to -1.
465 This parameter can be used to control the NMI watchdog
466 (i.e. the hard lockup detector) on x86 systems.
468 = =================================
469 0 Disable the hard lockup detector.
470 1 Enable the hard lockup detector.
471 = =================================
473 The hard lockup detector monitors each CPU for its ability to respond to
474 timer interrupts. The mechanism utilizes CPU performance counter registers
475 that are programmed to generate Non-Maskable Interrupts (NMIs) periodically
476 while a CPU is busy. Hence, the alternative name 'NMI watchdog'.
478 The NMI watchdog is disabled by default if the kernel is running as a guest
479 in a KVM virtual machine. This default can be overridden by adding::
483 to the guest kernel command line (see :doc:`/admin-guide/kernel-parameters`).
489 Enables/disables automatic page fault based NUMA memory
490 balancing. Memory is moved automatically to nodes
491 that access it often.
493 Enables/disables automatic NUMA memory balancing. On NUMA machines, there
494 is a performance penalty if remote memory is accessed by a CPU. When this
495 feature is enabled the kernel samples what task thread is accessing memory
496 by periodically unmapping pages and later trapping a page fault. At the
497 time of the page fault, it is determined if the data being accessed should
498 be migrated to a local memory node.
500 The unmapping of pages and trapping faults incur additional overhead that
501 ideally is offset by improved memory locality but there is no universal
502 guarantee. If the target workload is already bound to NUMA nodes then this
503 feature should be disabled. Otherwise, if the system overhead from the
504 feature is too high then the rate the kernel samples for NUMA hinting
505 faults may be controlled by the `numa_balancing_scan_period_min_ms,
506 numa_balancing_scan_delay_ms, numa_balancing_scan_period_max_ms,
507 numa_balancing_scan_size_mb`_, and numa_balancing_settle_count sysctls.
510 numa_balancing_scan_period_min_ms, numa_balancing_scan_delay_ms, numa_balancing_scan_period_max_ms, numa_balancing_scan_size_mb
511 ===============================================================================================================================
514 Automatic NUMA balancing scans tasks address space and unmaps pages to
515 detect if pages are properly placed or if the data should be migrated to a
516 memory node local to where the task is running. Every "scan delay" the task
517 scans the next "scan size" number of pages in its address space. When the
518 end of the address space is reached the scanner restarts from the beginning.
520 In combination, the "scan delay" and "scan size" determine the scan rate.
521 When "scan delay" decreases, the scan rate increases. The scan delay and
522 hence the scan rate of every task is adaptive and depends on historical
523 behaviour. If pages are properly placed then the scan delay increases,
524 otherwise the scan delay decreases. The "scan size" is not adaptive but
525 the higher the "scan size", the higher the scan rate.
527 Higher scan rates incur higher system overhead as page faults must be
528 trapped and potentially data must be migrated. However, the higher the scan
529 rate, the more quickly a tasks memory is migrated to a local node if the
530 workload pattern changes and minimises performance impact due to remote
531 memory accesses. These sysctls control the thresholds for scan delays and
532 the number of pages scanned.
534 ``numa_balancing_scan_period_min_ms`` is the minimum time in milliseconds to
535 scan a tasks virtual memory. It effectively controls the maximum scanning
538 ``numa_balancing_scan_delay_ms`` is the starting "scan delay" used for a task
539 when it initially forks.
541 ``numa_balancing_scan_period_max_ms`` is the maximum time in milliseconds to
542 scan a tasks virtual memory. It effectively controls the minimum scanning
545 ``numa_balancing_scan_size_mb`` is how many megabytes worth of pages are
546 scanned for a given scan.
549 osrelease, ostype & version
550 ===========================
559 #5 Wed Feb 25 21:49:24 MET 1998
561 The files ``osrelease`` and ``ostype`` should be clear enough.
563 needs a little more clarification however. The '#5' means that
564 this is the fifth kernel built from this source base and the
565 date behind it indicates the time the kernel was built.
566 The only way to tune these values is to rebuild the kernel :-)
569 overflowgid & overflowuid
570 =========================
572 if your architecture did not always support 32-bit UIDs (i.e. arm,
573 i386, m68k, sh, and sparc32), a fixed UID and GID will be returned to
574 applications that use the old 16-bit UID/GID system calls, if the
575 actual UID or GID would exceed 65535.
577 These sysctls allow you to change the value of the fixed UID and GID.
578 The default is 65534.
584 The value in this file determines the behaviour of the kernel on a
587 * if zero, the kernel will loop forever;
588 * if negative, the kernel will reboot immediately;
589 * if positive, the kernel will reboot after the corresponding number
592 When you use the software watchdog, the recommended setting is 60.
598 Controls the kernel's behavior when a CPU receives an NMI caused by
601 = ==================================================================
602 0 Try to continue operation (default).
603 1 Panic immediately. The IO error triggered an NMI. This indicates a
604 serious system condition which could result in IO data corruption.
605 Rather than continuing, panicking might be a better choice. Some
606 servers issue this sort of NMI when the dump button is pushed,
607 and you can use this option to take a crash dump.
608 = ==================================================================
614 Controls the kernel's behaviour when an oops or BUG is encountered.
616 = ===================================================================
617 0 Try to continue operation.
618 1 Panic immediately. If the `panic` sysctl is also non-zero then the
619 machine will be rebooted.
620 = ===================================================================
623 panic_on_stackoverflow
624 ======================
626 Controls the kernel's behavior when detecting the overflows of
627 kernel, IRQ and exception stacks except a user stack.
628 This file shows up if ``CONFIG_DEBUG_STACKOVERFLOW`` is enabled.
630 = ==========================
631 0 Try to continue operation.
633 = ==========================
636 panic_on_unrecovered_nmi
637 ========================
639 The default Linux behaviour on an NMI of either memory or unknown is
640 to continue operation. For many environments such as scientific
641 computing it is preferable that the box is taken out and the error
642 dealt with than an uncorrected parity/ECC error get propagated.
644 A small number of systems do generate NMIs for bizarre random reasons
645 such as power management so the default is off. That sysctl works like
646 the existing panic controls already in that directory.
652 Calls panic() in the WARN() path when set to 1. This is useful to avoid
653 a kernel rebuild when attempting to kdump at the location of a WARN().
655 = ================================================
656 0 Only WARN(), default behaviour.
657 1 Call panic() after printing out WARN() location.
658 = ================================================
664 Bitmask for printing system info when panic happens. User can chose
665 combination of the following bits:
667 ===== ============================================
668 bit 0 print all tasks info
669 bit 1 print system memory info
670 bit 2 print timer info
671 bit 3 print locks info if ``CONFIG_LOCKDEP`` is on
672 bit 4 print ftrace buffer
673 ===== ============================================
675 So for example to print tasks and memory info on panic, user can::
677 echo 3 > /proc/sys/kernel/panic_print
683 When set to 1, calls panic() after RCU stall detection messages. This
684 is useful to define the root cause of RCU stalls using a vmcore.
686 = ============================================================
687 0 Do not panic() when RCU stall takes place, default behavior.
688 1 panic() after printing RCU stall messages.
689 = ============================================================
692 perf_cpu_time_max_percent
693 =========================
695 Hints to the kernel how much CPU time it should be allowed to
696 use to handle perf sampling events. If the perf subsystem
697 is informed that its samples are exceeding this limit, it
698 will drop its sampling frequency to attempt to reduce its CPU
701 Some perf sampling happens in NMIs. If these samples
702 unexpectedly take too long to execute, the NMIs can become
703 stacked up next to each other so much that nothing else is
706 ===== ========================================================
707 0 Disable the mechanism. Do not monitor or correct perf's
708 sampling rate no matter how CPU time it takes.
710 1-100 Attempt to throttle perf's sample rate to this
711 percentage of CPU. Note: the kernel calculates an
712 "expected" length of each sample event. 100 here means
713 100% of that expected length. Even if this is set to
714 100, you may still see sample throttling if this
715 length is exceeded. Set to 0 if you truly do not care
716 how much CPU is consumed.
717 ===== ========================================================
723 Controls use of the performance events system by unprivileged
724 users (without CAP_PERFMON). The default value is 2.
726 For backward compatibility reasons access to system performance
727 monitoring and observability remains open for CAP_SYS_ADMIN
728 privileged processes but CAP_SYS_ADMIN usage for secure system
729 performance monitoring and observability operations is discouraged
730 with respect to CAP_PERFMON use cases.
732 === ==================================================================
733 -1 Allow use of (almost) all events by all users.
735 Ignore mlock limit after perf_event_mlock_kb without
738 >=0 Disallow ftrace function tracepoint by users without
741 Disallow raw tracepoint access by users without ``CAP_PERFMON``.
743 >=1 Disallow CPU event access by users without ``CAP_PERFMON``.
745 >=2 Disallow kernel profiling by users without ``CAP_PERFMON``.
746 === ==================================================================
752 Controls maximum number of stack frames to copy for (``attr.sample_type &
753 PERF_SAMPLE_CALLCHAIN``) configured events, for instance, when using
754 '``perf record -g``' or '``perf trace --call-graph fp``'.
756 This can only be done when no events are in use that have callchains
757 enabled, otherwise writing to this file will return ``-EBUSY``.
759 The default value is 127.
765 Control size of per-cpu ring buffer not counted agains mlock limit.
767 The default value is 512 + 1 page
770 perf_event_max_contexts_per_stack
771 =================================
773 Controls maximum number of stack frame context entries for
774 (``attr.sample_type & PERF_SAMPLE_CALLCHAIN``) configured events, for
775 instance, when using '``perf record -g``' or '``perf trace --call-graph fp``'.
777 This can only be done when no events are in use that have callchains
778 enabled, otherwise writing to this file will return ``-EBUSY``.
780 The default value is 8.
786 PID allocation wrap value. When the kernel's next PID value
787 reaches this value, it wraps back to a minimum PID value.
788 PIDs of value ``pid_max`` or larger are not allocated.
794 The last pid allocated in the current (the one task using this sysctl
795 lives in) pid namespace. When selecting a pid for a next task on fork
796 kernel tries to allocate a number starting from this one.
799 powersave-nap (PPC only)
800 ========================
802 If set, Linux-PPC will use the 'nap' mode of powersaving,
803 otherwise the 'doze' mode will be used.
806 ==============================================================
811 The four values in printk denote: ``console_loglevel``,
812 ``default_message_loglevel``, ``minimum_console_loglevel`` and
813 ``default_console_loglevel`` respectively.
815 These values influence printk() behavior when printing or
816 logging error messages. See '``man 2 syslog``' for more info on
817 the different loglevels.
819 ======================== =====================================
820 console_loglevel messages with a higher priority than
821 this will be printed to the console
822 default_message_loglevel messages without an explicit priority
823 will be printed with this priority
824 minimum_console_loglevel minimum (highest) value to which
825 console_loglevel can be set
826 default_console_loglevel default value for console_loglevel
827 ======================== =====================================
833 Delay each printk message in ``printk_delay`` milliseconds
835 Value from 0 - 10000 is allowed.
841 Some warning messages are rate limited. ``printk_ratelimit`` specifies
842 the minimum length of time between these messages (in seconds).
843 The default value is 5 seconds.
845 A value of 0 will disable rate limiting.
848 printk_ratelimit_burst
849 ======================
851 While long term we enforce one message per `printk_ratelimit`_
852 seconds, we do allow a burst of messages to pass through.
853 ``printk_ratelimit_burst`` specifies the number of messages we can
854 send before ratelimiting kicks in.
856 The default value is 10 messages.
862 Control the logging to ``/dev/kmsg`` from userspace:
864 ========= =============================================
865 ratelimit default, ratelimited
866 on unlimited logging to /dev/kmsg from userspace
867 off logging to /dev/kmsg disabled
868 ========= =============================================
870 The kernel command line parameter ``printk.devkmsg=`` overrides this and is
871 a one-time setting until next reboot: once set, it cannot be changed by
872 this sysctl interface anymore.
874 ==============================================================
880 See Documentation/filesystems/devpts.txt.
886 This option can be used to select the type of process address
887 space randomization that is used in the system, for architectures
888 that support this feature.
890 == ===========================================================================
891 0 Turn the process address space randomization off. This is the
892 default for architectures that do not support this feature anyways,
893 and kernels that are booted with the "norandmaps" parameter.
895 1 Make the addresses of mmap base, stack and VDSO page randomized.
896 This, among other things, implies that shared libraries will be
897 loaded to random addresses. Also for PIE-linked binaries, the
898 location of code start is randomized. This is the default if the
899 ``CONFIG_COMPAT_BRK`` option is enabled.
901 2 Additionally enable heap randomization. This is the default if
902 ``CONFIG_COMPAT_BRK`` is disabled.
904 There are a few legacy applications out there (such as some ancient
905 versions of libc.so.5 from 1996) that assume that brk area starts
906 just after the end of the code+bss. These applications break when
907 start of the brk area is randomized. There are however no known
908 non-legacy applications that would be broken this way, so for most
909 systems it is safe to choose full randomization.
911 Systems with ancient and/or broken binaries should be configured
912 with ``CONFIG_COMPAT_BRK`` enabled, which excludes the heap from process
913 address space randomization.
914 == ===========================================================================
920 See :doc:`/admin-guide/initrd`.
923 reboot-cmd (SPARC only)
924 =======================
926 ??? This seems to be a way to give an argument to the Sparc
927 ROM/Flash boot loader. Maybe to tell it what to do after
934 Enables/disables Energy Aware Scheduling (EAS). EAS starts
935 automatically on platforms where it can run (that is,
936 platforms with asymmetric CPU topologies and having an Energy
937 Model available). If your platform happens to meet the
938 requirements for EAS but you do not want to use it, change
945 Enables/disables scheduler statistics. Enabling this feature
946 incurs a small amount of overhead in the scheduler but is
947 useful for debugging and performance tuning.
953 See :doc:`/userspace-api/seccomp_filter`.
959 This file shows the size of the generic SCSI (sg) buffer.
960 You can't tune it just yet, but you could change it on
961 compile time by editing ``include/scsi/sg.h`` and changing
962 the value of ``SG_BIG_BUFF``.
964 There shouldn't be any reason to change this value. If
965 you can come up with one, you probably know what you
972 This parameter sets the total amount of shared memory pages that
973 can be used system wide. Hence, ``shmall`` should always be at least
974 ``ceil(shmmax/PAGE_SIZE)``.
976 If you are not sure what the default ``PAGE_SIZE`` is on your Linux
977 system, you can run the following command::
985 This value can be used to query and set the run time limit
986 on the maximum shared memory segment size that can be created.
987 Shared memory segments up to 1Gb are now supported in the
988 kernel. This value defaults to ``SHMMAX``.
994 This value determines the maximum number of shared memory segments.
995 4096 by default (``SHMMNI``).
1001 Linux lets you set resource limits, including how much memory one
1002 process can consume, via ``setrlimit(2)``. Unfortunately, shared memory
1003 segments are allowed to exist without association with any process, and
1004 thus might not be counted against any resource limits. If enabled,
1005 shared memory segments are automatically destroyed when their attach
1006 count becomes zero after a detach or a process termination. It will
1007 also destroy segments that were created, but never attached to, on exit
1008 from the process. The only use left for ``IPC_RMID`` is to immediately
1009 destroy an unattached segment. Of course, this breaks the way things are
1010 defined, so some applications might stop working. Note that this
1011 feature will do you no good unless you also configure your resource
1012 limits (in particular, ``RLIMIT_AS`` and ``RLIMIT_NPROC``). Most systems don't
1015 Note that if you change this from 0 to 1, already created segments
1016 without users and with a dead originative process will be destroyed.
1019 sysctl_writes_strict
1020 ====================
1022 Control how file position affects the behavior of updating sysctl values
1023 via the ``/proc/sys`` interface:
1025 == ======================================================================
1026 -1 Legacy per-write sysctl value handling, with no printk warnings.
1027 Each write syscall must fully contain the sysctl value to be
1028 written, and multiple writes on the same sysctl file descriptor
1029 will rewrite the sysctl value, regardless of file position.
1030 0 Same behavior as above, but warn about processes that perform writes
1031 to a sysctl file descriptor when the file position is not 0.
1032 1 (default) Respect file position when writing sysctl strings. Multiple
1033 writes will append to the sysctl value buffer. Anything past the max
1034 length of the sysctl value buffer will be ignored. Writes to numeric
1035 sysctl entries must always be at file position 0 and the value must
1036 be fully contained in the buffer sent in the write syscall.
1037 == ======================================================================
1040 softlockup_all_cpu_backtrace
1041 ============================
1043 This value controls the soft lockup detector thread's behavior
1044 when a soft lockup condition is detected as to whether or not
1045 to gather further debug information. If enabled, each cpu will
1046 be issued an NMI and instructed to capture stack trace.
1048 This feature is only applicable for architectures which support
1051 = ============================================
1052 0 Do nothing. This is the default behavior.
1053 1 On detection capture more debug information.
1054 = ============================================
1060 This parameter can be used to control whether the kernel panics
1061 when a soft lockup is detected.
1063 = ============================================
1064 0 Don't panic on soft lockup.
1065 1 Panic on soft lockup.
1066 = ============================================
1068 This can also be set using the softlockup_panic kernel parameter.
1074 This parameter can be used to control the soft lockup detector.
1076 = =================================
1077 0 Disable the soft lockup detector.
1078 1 Enable the soft lockup detector.
1079 = =================================
1081 The soft lockup detector monitors CPUs for threads that are hogging the CPUs
1082 without rescheduling voluntarily, and thus prevent the 'watchdog/N' threads
1083 from running. The mechanism depends on the CPUs ability to respond to timer
1084 interrupts which are needed for the 'watchdog/N' threads to be woken up by
1085 the watchdog timer function, otherwise the NMI watchdog — if enabled — can
1086 detect a hard lockup condition.
1092 This parameter can be used to control kernel stack erasing at the end
1093 of syscalls for kernels built with ``CONFIG_GCC_PLUGIN_STACKLEAK``.
1095 That erasing reduces the information which kernel stack leak bugs
1096 can reveal and blocks some uninitialized stack variable attacks.
1097 The tradeoff is the performance impact: on a single CPU system kernel
1098 compilation sees a 1% slowdown, other systems and workloads may vary.
1100 = ====================================================================
1101 0 Kernel stack erasing is disabled, STACKLEAK_METRICS are not updated.
1102 1 Kernel stack erasing is enabled (default), it is performed before
1103 returning to the userspace at the end of syscalls.
1104 = ====================================================================
1112 = ====================================
1113 0 Stop-A has no effect.
1114 1 Stop-A breaks to the PROM (default).
1115 = ====================================
1117 Stop-A is always enabled on a panic, so that the user can return to
1124 See :doc:`/admin-guide/sysrq`.
1130 Non-zero if the kernel has been tainted. Numeric values, which can be
1131 ORed together. The letters are seen in "Tainted" line of Oops reports.
1133 ====== ===== ==============================================================
1134 1 `(P)` proprietary module was loaded
1135 2 `(F)` module was force loaded
1136 4 `(S)` SMP kernel oops on an officially SMP incapable processor
1137 8 `(R)` module was force unloaded
1138 16 `(M)` processor reported a Machine Check Exception (MCE)
1139 32 `(B)` bad page referenced or some unexpected page flags
1140 64 `(U)` taint requested by userspace application
1141 128 `(D)` kernel died recently, i.e. there was an OOPS or BUG
1142 256 `(A)` an ACPI table was overridden by user
1143 512 `(W)` kernel issued warning
1144 1024 `(C)` staging driver was loaded
1145 2048 `(I)` workaround for bug in platform firmware applied
1146 4096 `(O)` externally-built ("out-of-tree") module was loaded
1147 8192 `(E)` unsigned module was loaded
1148 16384 `(L)` soft lockup occurred
1149 32768 `(K)` kernel has been live patched
1150 65536 `(X)` Auxiliary taint, defined and used by for distros
1151 131072 `(T)` The kernel was built with the struct randomization plugin
1152 ====== ===== ==============================================================
1154 See :doc:`/admin-guide/tainted-kernels` for more information.
1160 This value controls the maximum number of threads that can be created
1163 During initialization the kernel sets this value such that even if the
1164 maximum number of threads is created, the thread structures occupy only
1165 a part (1/8th) of the available RAM pages.
1167 The minimum value that can be written to ``threads-max`` is 1.
1169 The maximum value that can be written to ``threads-max`` is given by the
1170 constant ``FUTEX_TID_MASK`` (0x3fffffff).
1172 If a value outside of this range is written to ``threads-max`` an
1173 ``EINVAL`` error occurs.
1179 The value in this file affects behavior of handling NMI. When the
1180 value is non-zero, unknown NMI is trapped and then panic occurs. At
1181 that time, kernel debugging information is displayed on console.
1183 NMI switch that most IA32 servers have fires unknown NMI up, for
1184 example. If a system hangs up, try pressing the NMI switch.
1190 This parameter can be used to disable or enable the soft lockup detector
1191 *and* the NMI watchdog (i.e. the hard lockup detector) at the same time.
1193 = ==============================
1194 0 Disable both lockup detectors.
1195 1 Enable both lockup detectors.
1196 = ==============================
1198 The soft lockup detector and the NMI watchdog can also be disabled or
1199 enabled individually, using the ``soft_watchdog`` and ``nmi_watchdog``
1201 If the ``watchdog`` parameter is read, for example by executing::
1203 cat /proc/sys/kernel/watchdog
1205 the output of this command (0 or 1) shows the logical OR of
1206 ``soft_watchdog`` and ``nmi_watchdog``.
1212 This value can be used to control on which cpus the watchdog may run.
1213 The default cpumask is all possible cores, but if ``NO_HZ_FULL`` is
1214 enabled in the kernel config, and cores are specified with the
1215 ``nohz_full=`` boot argument, those cores are excluded by default.
1216 Offline cores can be included in this mask, and if the core is later
1217 brought online, the watchdog will be started based on the mask value.
1219 Typically this value would only be touched in the ``nohz_full`` case
1220 to re-enable cores that by default were not running the watchdog,
1221 if a kernel lockup was suspected on those cores.
1223 The argument value is the standard cpulist format for cpumasks,
1224 so for example to enable the watchdog on cores 0, 2, 3, and 4 you
1227 echo 0,2-4 > /proc/sys/kernel/watchdog_cpumask
1233 This value can be used to control the frequency of hrtimer and NMI
1234 events and the soft and hard lockup thresholds. The default threshold
1237 The softlockup threshold is (``2 * watchdog_thresh``). Setting this
1238 tunable to zero will disable lockup detection altogether.