1 Documentation 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>
5 For general info and legal blurb, please look in README.
7 ==============================================================
9 This file contains documentation for the sysctl files in
10 /proc/sys/kernel/ and is valid for Linux kernel version 2.2.
12 The files in this directory can be used to tune and monitor
13 miscellaneous and general things in the operation of the Linux
14 kernel. Since some of the files _can_ be used to screw up your
15 system, it is advisable to read both documentation and source
16 before actually making adjustments.
18 Currently, these files might (depending on your configuration)
19 show up in /proc/sys/kernel:
24 - bootloader_type [ X86 only ]
25 - bootloader_version [ X86 only ]
26 - callhome [ S390 only ]
36 - hardlockup_all_cpu_backtrace
38 - hung_task_check_count
39 - hung_task_timeout_secs
44 - modprobe ==> Documentation/debugging-modules.txt
46 - msg_next_id [ sysv ipc ]
58 - panic_on_stackoverflow
59 - panic_on_unrecovered_nmi
62 - perf_cpu_time_max_percent
64 - perf_event_max_stack
66 - perf_event_max_contexts_per_stack
68 - powersave-nap [ PPC only ]
72 - printk_ratelimit_burst
73 - pty ==> Documentation/filesystems/devpts.txt
75 - real-root-dev ==> Documentation/admin-guide/initrd.rst
76 - reboot-cmd [ SPARC only ]
79 - seccomp/ ==> Documentation/userspace-api/seccomp_filter.rst
81 - sem_next_id [ sysv ipc ]
82 - sg-big-buff [ generic SCSI device (sg) ]
83 - shm_next_id [ sysv ipc ]
88 - softlockup_all_cpu_backtrace
90 - stop-a [ SPARC only ]
91 - sysrq ==> Documentation/admin-guide/sysrq.rst
92 - sysctl_writes_strict
95 - unprivileged_bpf_disabled
102 ==============================================================
106 highwater lowwater frequency
108 If BSD-style process accounting is enabled these values control
109 its behaviour. If free space on filesystem where the log lives
110 goes below <lowwater>% accounting suspends. If free space gets
111 above <highwater>% accounting resumes. <Frequency> determines
112 how often do we check the amount of free space (value is in
115 That is, suspend accounting if there left <= 2% free; resume it
116 if we got >=4%; consider information about amount of free space
117 valid for 30 seconds.
119 ==============================================================
125 See Doc*/kernel/power/video.txt, it allows mode of video boot to be
128 ==============================================================
132 This variable has no effect and may be removed in future kernel
133 releases. Reading it always returns 0.
134 Up to Linux 3.17, it enabled/disabled automatic recomputing of msgmni
135 upon memory add/remove or upon ipc namespace creation/removal.
136 Echoing "1" into this file enabled msgmni automatic recomputing.
137 Echoing "0" turned it off. auto_msgmni default value was 1.
140 ==============================================================
144 x86 bootloader identification
146 This gives the bootloader type number as indicated by the bootloader,
147 shifted left by 4, and OR'd with the low four bits of the bootloader
148 version. The reason for this encoding is that this used to match the
149 type_of_loader field in the kernel header; the encoding is kept for
150 backwards compatibility. That is, if the full bootloader type number
151 is 0x15 and the full version number is 0x234, this file will contain
152 the value 340 = 0x154.
154 See the type_of_loader and ext_loader_type fields in
155 Documentation/x86/boot.txt for additional information.
157 ==============================================================
161 x86 bootloader version
163 The complete bootloader version number. In the example above, this
164 file will contain the value 564 = 0x234.
166 See the type_of_loader and ext_loader_ver fields in
167 Documentation/x86/boot.txt for additional information.
169 ==============================================================
173 Controls the kernel's callhome behavior in case of a kernel panic.
175 The s390 hardware allows an operating system to send a notification
176 to a service organization (callhome) in case of an operating system panic.
178 When the value in this file is 0 (which is the default behavior)
179 nothing happens in case of a kernel panic. If this value is set to "1"
180 the complete kernel oops message is send to the IBM customer service
181 organization in case the mainframe the Linux operating system is running
182 on has a service contract with IBM.
184 ==============================================================
188 Highest valid capability of the running kernel. Exports
189 CAP_LAST_CAP from the kernel.
191 ==============================================================
195 core_pattern is used to specify a core dumpfile pattern name.
196 . max length 128 characters; default value is "core"
197 . core_pattern is used as a pattern template for the output filename;
198 certain string patterns (beginning with '%') are substituted with
200 . backward compatibility with core_uses_pid:
201 If core_pattern does not include "%p" (default does not)
202 and core_uses_pid is set, then .PID will be appended to
204 . corename format specifiers:
205 %<NUL> '%' is dropped
208 %P global pid (init PID namespace)
210 %I global tid (init PID namespace)
211 %u uid (in initial user namespace)
212 %g gid (in initial user namespace)
213 %d dump mode, matches PR_SET_DUMPABLE and
214 /proc/sys/fs/suid_dumpable
218 %e executable filename (may be shortened)
220 %<OTHER> both are dropped
221 . If the first character of the pattern is a '|', the kernel will treat
222 the rest of the pattern as a command to run. The core dump will be
223 written to the standard input of that program instead of to a file.
225 ==============================================================
229 This sysctl is only applicable when core_pattern is configured to pipe
230 core files to a user space helper (when the first character of
231 core_pattern is a '|', see above). When collecting cores via a pipe
232 to an application, it is occasionally useful for the collecting
233 application to gather data about the crashing process from its
234 /proc/pid directory. In order to do this safely, the kernel must wait
235 for the collecting process to exit, so as not to remove the crashing
236 processes proc files prematurely. This in turn creates the
237 possibility that a misbehaving userspace collecting process can block
238 the reaping of a crashed process simply by never exiting. This sysctl
239 defends against that. It defines how many concurrent crashing
240 processes may be piped to user space applications in parallel. If
241 this value is exceeded, then those crashing processes above that value
242 are noted via the kernel log and their cores are skipped. 0 is a
243 special value, indicating that unlimited processes may be captured in
244 parallel, but that no waiting will take place (i.e. the collecting
245 process is not guaranteed access to /proc/<crashing pid>/). This
248 ==============================================================
252 The default coredump filename is "core". By setting
253 core_uses_pid to 1, the coredump filename becomes core.PID.
254 If core_pattern does not include "%p" (default does not)
255 and core_uses_pid is set, then .PID will be appended to
258 ==============================================================
262 When the value in this file is 0, ctrl-alt-del is trapped and
263 sent to the init(1) program to handle a graceful restart.
264 When, however, the value is > 0, Linux's reaction to a Vulcan
265 Nerve Pinch (tm) will be an immediate reboot, without even
266 syncing its dirty buffers.
268 Note: when a program (like dosemu) has the keyboard in 'raw'
269 mode, the ctrl-alt-del is intercepted by the program before it
270 ever reaches the kernel tty layer, and it's up to the program
271 to decide what to do with it.
273 ==============================================================
277 This toggle indicates whether unprivileged users are prevented
278 from using dmesg(8) to view messages from the kernel's log buffer.
279 When dmesg_restrict is set to (0) there are no restrictions. When
280 dmesg_restrict is set set to (1), users must have CAP_SYSLOG to use
283 The kernel config option CONFIG_SECURITY_DMESG_RESTRICT sets the
284 default value of dmesg_restrict.
286 ==============================================================
288 domainname & hostname:
290 These files can be used to set the NIS/YP domainname and the
291 hostname of your box in exactly the same way as the commands
292 domainname and hostname, i.e.:
293 # echo "darkstar" > /proc/sys/kernel/hostname
294 # echo "mydomain" > /proc/sys/kernel/domainname
295 has the same effect as
296 # hostname "darkstar"
297 # domainname "mydomain"
299 Note, however, that the classic darkstar.frop.org has the
300 hostname "darkstar" and DNS (Internet Domain Name Server)
301 domainname "frop.org", not to be confused with the NIS (Network
302 Information Service) or YP (Yellow Pages) domainname. These two
303 domain names are in general different. For a detailed discussion
304 see the hostname(1) man page.
306 ==============================================================
307 hardlockup_all_cpu_backtrace:
309 This value controls the hard lockup detector behavior when a hard
310 lockup condition is detected as to whether or not to gather further
311 debug information. If enabled, arch-specific all-CPU stack dumping
314 0: do nothing. This is the default behavior.
316 1: on detection capture more debug information.
317 ==============================================================
321 Path for the hotplug policy agent.
322 Default value is "/sbin/hotplug".
324 ==============================================================
328 Controls the kernel's behavior when a hung task is detected.
329 This file shows up if CONFIG_DETECT_HUNG_TASK is enabled.
331 0: continue operation. This is the default behavior.
333 1: panic immediately.
335 ==============================================================
337 hung_task_check_count:
339 The upper bound on the number of tasks that are checked.
340 This file shows up if CONFIG_DETECT_HUNG_TASK is enabled.
342 ==============================================================
344 hung_task_timeout_secs:
346 Check interval. When a task in D state did not get scheduled
347 for more than this value report a warning.
348 This file shows up if CONFIG_DETECT_HUNG_TASK is enabled.
350 0: means infinite timeout - no checking done.
351 Possible values to set are in range {0..LONG_MAX/HZ}.
353 ==============================================================
357 The maximum number of warnings to report. During a check interval
358 if a hung task is detected, this value is decreased by 1.
359 When this value reaches 0, no more warnings will be reported.
360 This file shows up if CONFIG_DETECT_HUNG_TASK is enabled.
362 -1: report an infinite number of warnings.
364 ==============================================================
368 A toggle indicating if the kexec_load syscall has been disabled. This
369 value defaults to 0 (false: kexec_load enabled), but can be set to 1
370 (true: kexec_load disabled). Once true, kexec can no longer be used, and
371 the toggle cannot be set back to false. This allows a kexec image to be
372 loaded before disabling the syscall, allowing a system to set up (and
373 later use) an image without it being altered. Generally used together
374 with the "modules_disabled" sysctl.
376 ==============================================================
380 This toggle indicates whether restrictions are placed on
381 exposing kernel addresses via /proc and other interfaces.
383 When kptr_restrict is set to (0), the default, there are no restrictions.
385 When kptr_restrict is set to (1), kernel pointers printed using the %pK
386 format specifier will be replaced with 0's unless the user has CAP_SYSLOG
387 and effective user and group ids are equal to the real ids. This is
388 because %pK checks are done at read() time rather than open() time, so
389 if permissions are elevated between the open() and the read() (e.g via
390 a setuid binary) then %pK will not leak kernel pointers to unprivileged
391 users. Note, this is a temporary solution only. The correct long-term
392 solution is to do the permission checks at open() time. Consider removing
393 world read permissions from files that use %pK, and using dmesg_restrict
394 to protect against uses of %pK in dmesg(8) if leaking kernel pointer
395 values to unprivileged users is a concern.
397 When kptr_restrict is set to (2), kernel pointers printed using
398 %pK will be replaced with 0's regardless of privileges.
400 ==============================================================
404 This flag controls the L2 cache of G3 processor boards. If
405 0, the cache is disabled. Enabled if nonzero.
407 ==============================================================
411 A toggle value indicating if modules are allowed to be loaded
412 in an otherwise modular kernel. This toggle defaults to off
413 (0), but can be set true (1). Once true, modules can be
414 neither loaded nor unloaded, and the toggle cannot be set back
415 to false. Generally used with the "kexec_load_disabled" toggle.
417 ==============================================================
419 msg_next_id, sem_next_id, and shm_next_id:
421 These three toggles allows to specify desired id for next allocated IPC
422 object: message, semaphore or shared memory respectively.
424 By default they are equal to -1, which means generic allocation logic.
425 Possible values to set are in range {0..INT_MAX}.
428 1) kernel doesn't guarantee, that new object will have desired id. So,
429 it's up to userspace, how to handle an object with "wrong" id.
430 2) Toggle with non-default value will be set back to -1 by kernel after
431 successful IPC object allocation.
433 ==============================================================
437 This parameter can be used to control the NMI watchdog
438 (i.e. the hard lockup detector) on x86 systems.
440 0 - disable the hard lockup detector
441 1 - enable the hard lockup detector
443 The hard lockup detector monitors each CPU for its ability to respond to
444 timer interrupts. The mechanism utilizes CPU performance counter registers
445 that are programmed to generate Non-Maskable Interrupts (NMIs) periodically
446 while a CPU is busy. Hence, the alternative name 'NMI watchdog'.
448 The NMI watchdog is disabled by default if the kernel is running as a guest
449 in a KVM virtual machine. This default can be overridden by adding
453 to the guest kernel command line (see Documentation/admin-guide/kernel-parameters.rst).
455 ==============================================================
459 Enables/disables automatic page fault based NUMA memory
460 balancing. Memory is moved automatically to nodes
461 that access it often.
463 Enables/disables automatic NUMA memory balancing. On NUMA machines, there
464 is a performance penalty if remote memory is accessed by a CPU. When this
465 feature is enabled the kernel samples what task thread is accessing memory
466 by periodically unmapping pages and later trapping a page fault. At the
467 time of the page fault, it is determined if the data being accessed should
468 be migrated to a local memory node.
470 The unmapping of pages and trapping faults incur additional overhead that
471 ideally is offset by improved memory locality but there is no universal
472 guarantee. If the target workload is already bound to NUMA nodes then this
473 feature should be disabled. Otherwise, if the system overhead from the
474 feature is too high then the rate the kernel samples for NUMA hinting
475 faults may be controlled by the numa_balancing_scan_period_min_ms,
476 numa_balancing_scan_delay_ms, numa_balancing_scan_period_max_ms,
477 numa_balancing_scan_size_mb, and numa_balancing_settle_count sysctls.
479 ==============================================================
481 numa_balancing_scan_period_min_ms, numa_balancing_scan_delay_ms,
482 numa_balancing_scan_period_max_ms, numa_balancing_scan_size_mb
484 Automatic NUMA balancing scans tasks address space and unmaps pages to
485 detect if pages are properly placed or if the data should be migrated to a
486 memory node local to where the task is running. Every "scan delay" the task
487 scans the next "scan size" number of pages in its address space. When the
488 end of the address space is reached the scanner restarts from the beginning.
490 In combination, the "scan delay" and "scan size" determine the scan rate.
491 When "scan delay" decreases, the scan rate increases. The scan delay and
492 hence the scan rate of every task is adaptive and depends on historical
493 behaviour. If pages are properly placed then the scan delay increases,
494 otherwise the scan delay decreases. The "scan size" is not adaptive but
495 the higher the "scan size", the higher the scan rate.
497 Higher scan rates incur higher system overhead as page faults must be
498 trapped and potentially data must be migrated. However, the higher the scan
499 rate, the more quickly a tasks memory is migrated to a local node if the
500 workload pattern changes and minimises performance impact due to remote
501 memory accesses. These sysctls control the thresholds for scan delays and
502 the number of pages scanned.
504 numa_balancing_scan_period_min_ms is the minimum time in milliseconds to
505 scan a tasks virtual memory. It effectively controls the maximum scanning
508 numa_balancing_scan_delay_ms is the starting "scan delay" used for a task
509 when it initially forks.
511 numa_balancing_scan_period_max_ms is the maximum time in milliseconds to
512 scan a tasks virtual memory. It effectively controls the minimum scanning
515 numa_balancing_scan_size_mb is how many megabytes worth of pages are
516 scanned for a given scan.
518 ==============================================================
522 Number of kernel oopses after which the kernel should panic when
523 ``panic_on_oops`` is not set. Setting this to 0 disables checking
524 the count. Setting this to 1 has the same effect as setting
525 ``panic_on_oops=1``. The default value is 10000.
527 ==============================================================
529 osrelease, ostype & version:
536 #5 Wed Feb 25 21:49:24 MET 1998
538 The files osrelease and ostype should be clear enough. Version
539 needs a little more clarification however. The '#5' means that
540 this is the fifth kernel built from this source base and the
541 date behind it indicates the time the kernel was built.
542 The only way to tune these values is to rebuild the kernel :-)
544 ==============================================================
546 overflowgid & overflowuid:
548 if your architecture did not always support 32-bit UIDs (i.e. arm,
549 i386, m68k, sh, and sparc32), a fixed UID and GID will be returned to
550 applications that use the old 16-bit UID/GID system calls, if the
551 actual UID or GID would exceed 65535.
553 These sysctls allow you to change the value of the fixed UID and GID.
554 The default is 65534.
556 ==============================================================
560 The value in this file represents the number of seconds the kernel
561 waits before rebooting on a panic. When you use the software watchdog,
562 the recommended setting is 60.
564 ==============================================================
568 Controls the kernel's behavior when a CPU receives an NMI caused by
571 0: try to continue operation (default)
573 1: panic immediately. The IO error triggered an NMI. This indicates a
574 serious system condition which could result in IO data corruption.
575 Rather than continuing, panicking might be a better choice. Some
576 servers issue this sort of NMI when the dump button is pushed,
577 and you can use this option to take a crash dump.
579 ==============================================================
583 Controls the kernel's behaviour when an oops or BUG is encountered.
585 0: try to continue operation
587 1: panic immediately. If the `panic' sysctl is also non-zero then the
588 machine will be rebooted.
590 ==============================================================
592 panic_on_stackoverflow:
594 Controls the kernel's behavior when detecting the overflows of
595 kernel, IRQ and exception stacks except a user stack.
596 This file shows up if CONFIG_DEBUG_STACKOVERFLOW is enabled.
598 0: try to continue operation.
600 1: panic immediately.
602 ==============================================================
604 panic_on_unrecovered_nmi:
606 The default Linux behaviour on an NMI of either memory or unknown is
607 to continue operation. For many environments such as scientific
608 computing it is preferable that the box is taken out and the error
609 dealt with than an uncorrected parity/ECC error get propagated.
611 A small number of systems do generate NMI's for bizarre random reasons
612 such as power management so the default is off. That sysctl works like
613 the existing panic controls already in that directory.
615 ==============================================================
619 Calls panic() in the WARN() path when set to 1. This is useful to avoid
620 a kernel rebuild when attempting to kdump at the location of a WARN().
622 0: only WARN(), default behaviour.
624 1: call panic() after printing out WARN() location.
626 ==============================================================
630 When set to 1, calls panic() after RCU stall detection messages. This
631 is useful to define the root cause of RCU stalls using a vmcore.
633 0: do not panic() when RCU stall takes place, default behavior.
635 1: panic() after printing RCU stall messages.
637 ==============================================================
639 perf_cpu_time_max_percent:
641 Hints to the kernel how much CPU time it should be allowed to
642 use to handle perf sampling events. If the perf subsystem
643 is informed that its samples are exceeding this limit, it
644 will drop its sampling frequency to attempt to reduce its CPU
647 Some perf sampling happens in NMIs. If these samples
648 unexpectedly take too long to execute, the NMIs can become
649 stacked up next to each other so much that nothing else is
652 0: disable the mechanism. Do not monitor or correct perf's
653 sampling rate no matter how CPU time it takes.
655 1-100: attempt to throttle perf's sample rate to this
656 percentage of CPU. Note: the kernel calculates an
657 "expected" length of each sample event. 100 here means
658 100% of that expected length. Even if this is set to
659 100, you may still see sample throttling if this
660 length is exceeded. Set to 0 if you truly do not care
661 how much CPU is consumed.
663 ==============================================================
667 Controls use of the performance events system by unprivileged
668 users (without CAP_SYS_ADMIN). The default value is 2.
670 -1: Allow use of (almost) all events by all users
671 Ignore mlock limit after perf_event_mlock_kb without CAP_IPC_LOCK
672 >=0: Disallow ftrace function tracepoint by users without CAP_SYS_ADMIN
673 Disallow raw tracepoint access by users without CAP_SYS_ADMIN
674 >=1: Disallow CPU event access by users without CAP_SYS_ADMIN
675 >=2: Disallow kernel profiling by users without CAP_SYS_ADMIN
677 ==============================================================
679 perf_event_max_stack:
681 Controls maximum number of stack frames to copy for (attr.sample_type &
682 PERF_SAMPLE_CALLCHAIN) configured events, for instance, when using
683 'perf record -g' or 'perf trace --call-graph fp'.
685 This can only be done when no events are in use that have callchains
686 enabled, otherwise writing to this file will return -EBUSY.
688 The default value is 127.
690 ==============================================================
694 Control size of per-cpu ring buffer not counted agains mlock limit.
696 The default value is 512 + 1 page
698 ==============================================================
700 perf_event_max_contexts_per_stack:
702 Controls maximum number of stack frame context entries for
703 (attr.sample_type & PERF_SAMPLE_CALLCHAIN) configured events, for
704 instance, when using 'perf record -g' or 'perf trace --call-graph fp'.
706 This can only be done when no events are in use that have callchains
707 enabled, otherwise writing to this file will return -EBUSY.
709 The default value is 8.
711 ==============================================================
715 PID allocation wrap value. When the kernel's next PID value
716 reaches this value, it wraps back to a minimum PID value.
717 PIDs of value pid_max or larger are not allocated.
719 ==============================================================
723 The last pid allocated in the current (the one task using this sysctl
724 lives in) pid namespace. When selecting a pid for a next task on fork
725 kernel tries to allocate a number starting from this one.
727 ==============================================================
729 powersave-nap: (PPC only)
731 If set, Linux-PPC will use the 'nap' mode of powersaving,
732 otherwise the 'doze' mode will be used.
734 ==============================================================
738 The four values in printk denote: console_loglevel,
739 default_message_loglevel, minimum_console_loglevel and
740 default_console_loglevel respectively.
742 These values influence printk() behavior when printing or
743 logging error messages. See 'man 2 syslog' for more info on
744 the different loglevels.
746 - console_loglevel: messages with a higher priority than
747 this will be printed to the console
748 - default_message_loglevel: messages without an explicit priority
749 will be printed with this priority
750 - minimum_console_loglevel: minimum (highest) value to which
751 console_loglevel can be set
752 - default_console_loglevel: default value for console_loglevel
754 ==============================================================
758 Delay each printk message in printk_delay milliseconds
760 Value from 0 - 10000 is allowed.
762 ==============================================================
766 Some warning messages are rate limited. printk_ratelimit specifies
767 the minimum length of time between these messages (in jiffies), by
768 default we allow one every 5 seconds.
770 A value of 0 will disable rate limiting.
772 ==============================================================
774 printk_ratelimit_burst:
776 While long term we enforce one message per printk_ratelimit
777 seconds, we do allow a burst of messages to pass through.
778 printk_ratelimit_burst specifies the number of messages we can
779 send before ratelimiting kicks in.
781 ==============================================================
785 Control the logging to /dev/kmsg from userspace:
787 ratelimit: default, ratelimited
788 on: unlimited logging to /dev/kmsg from userspace
789 off: logging to /dev/kmsg disabled
791 The kernel command line parameter printk.devkmsg= overrides this and is
792 a one-time setting until next reboot: once set, it cannot be changed by
793 this sysctl interface anymore.
798 See Documentation/filesystems/devpts.rst.
804 This is a directory, with the following entries:
806 * ``boot_id``: a UUID generated the first time this is retrieved, and
807 unvarying after that;
809 * ``uuid``: a UUID generated every time this is retrieved (this can
810 thus be used to generate UUIDs at will);
812 * ``entropy_avail``: the pool's entropy count, in bits;
814 * ``poolsize``: the entropy pool size, in bits;
816 * ``urandom_min_reseed_secs``: obsolete (used to determine the minimum
817 number of seconds between urandom pool reseeding). This file is
818 writable for compatibility purposes, but writing to it has no effect
821 * ``write_wakeup_threshold``: when the entropy count drops below this
822 (as a number of bits), processes waiting to write to ``/dev/random``
823 are woken up. This file is writable for compatibility purposes, but
824 writing to it has no effect on any RNG behavior.
830 This option can be used to select the type of process address
831 space randomization that is used in the system, for architectures
832 that support this feature.
834 0 - Turn the process address space randomization off. This is the
835 default for architectures that do not support this feature anyways,
836 and kernels that are booted with the "norandmaps" parameter.
838 1 - Make the addresses of mmap base, stack and VDSO page randomized.
839 This, among other things, implies that shared libraries will be
840 loaded to random addresses. Also for PIE-linked binaries, the
841 location of code start is randomized. This is the default if the
842 CONFIG_COMPAT_BRK option is enabled.
844 2 - Additionally enable heap randomization. This is the default if
845 CONFIG_COMPAT_BRK is disabled.
847 There are a few legacy applications out there (such as some ancient
848 versions of libc.so.5 from 1996) that assume that brk area starts
849 just after the end of the code+bss. These applications break when
850 start of the brk area is randomized. There are however no known
851 non-legacy applications that would be broken this way, so for most
852 systems it is safe to choose full randomization.
854 Systems with ancient and/or broken binaries should be configured
855 with CONFIG_COMPAT_BRK enabled, which excludes the heap from process
856 address space randomization.
858 ==============================================================
860 reboot-cmd: (Sparc only)
862 ??? This seems to be a way to give an argument to the Sparc
863 ROM/Flash boot loader. Maybe to tell it what to do after
866 ==============================================================
868 rtsig-max & rtsig-nr:
870 The file rtsig-max can be used to tune the maximum number
871 of POSIX realtime (queued) signals that can be outstanding
874 rtsig-nr shows the number of RT signals currently queued.
876 ==============================================================
880 Enables/disables scheduler statistics. Enabling this feature
881 incurs a small amount of overhead in the scheduler but is
882 useful for debugging and performance tuning.
884 ==============================================================
888 This file shows the size of the generic SCSI (sg) buffer.
889 You can't tune it just yet, but you could change it on
890 compile time by editing include/scsi/sg.h and changing
891 the value of SG_BIG_BUFF.
893 There shouldn't be any reason to change this value. If
894 you can come up with one, you probably know what you
897 ==============================================================
901 This parameter sets the total amount of shared memory pages that
902 can be used system wide. Hence, SHMALL should always be at least
903 ceil(shmmax/PAGE_SIZE).
905 If you are not sure what the default PAGE_SIZE is on your Linux
906 system, you can run the following command:
910 ==============================================================
914 This value can be used to query and set the run time limit
915 on the maximum shared memory segment size that can be created.
916 Shared memory segments up to 1Gb are now supported in the
917 kernel. This value defaults to SHMMAX.
919 ==============================================================
923 Linux lets you set resource limits, including how much memory one
924 process can consume, via setrlimit(2). Unfortunately, shared memory
925 segments are allowed to exist without association with any process, and
926 thus might not be counted against any resource limits. If enabled,
927 shared memory segments are automatically destroyed when their attach
928 count becomes zero after a detach or a process termination. It will
929 also destroy segments that were created, but never attached to, on exit
930 from the process. The only use left for IPC_RMID is to immediately
931 destroy an unattached segment. Of course, this breaks the way things are
932 defined, so some applications might stop working. Note that this
933 feature will do you no good unless you also configure your resource
934 limits (in particular, RLIMIT_AS and RLIMIT_NPROC). Most systems don't
937 Note that if you change this from 0 to 1, already created segments
938 without users and with a dead originative process will be destroyed.
940 ==============================================================
942 sysctl_writes_strict:
944 Control how file position affects the behavior of updating sysctl values
945 via the /proc/sys interface:
947 -1 - Legacy per-write sysctl value handling, with no printk warnings.
948 Each write syscall must fully contain the sysctl value to be
949 written, and multiple writes on the same sysctl file descriptor
950 will rewrite the sysctl value, regardless of file position.
951 0 - Same behavior as above, but warn about processes that perform writes
952 to a sysctl file descriptor when the file position is not 0.
953 1 - (default) Respect file position when writing sysctl strings. Multiple
954 writes will append to the sysctl value buffer. Anything past the max
955 length of the sysctl value buffer will be ignored. Writes to numeric
956 sysctl entries must always be at file position 0 and the value must
957 be fully contained in the buffer sent in the write syscall.
959 ==============================================================
961 softlockup_all_cpu_backtrace:
963 This value controls the soft lockup detector thread's behavior
964 when a soft lockup condition is detected as to whether or not
965 to gather further debug information. If enabled, each cpu will
966 be issued an NMI and instructed to capture stack trace.
968 This feature is only applicable for architectures which support
971 0: do nothing. This is the default behavior.
973 1: on detection capture more debug information.
975 ==============================================================
979 This parameter can be used to control the soft lockup detector.
981 0 - disable the soft lockup detector
982 1 - enable the soft lockup detector
984 The soft lockup detector monitors CPUs for threads that are hogging the CPUs
985 without rescheduling voluntarily, and thus prevent the 'watchdog/N' threads
986 from running. The mechanism depends on the CPUs ability to respond to timer
987 interrupts which are needed for the 'watchdog/N' threads to be woken up by
988 the watchdog timer function, otherwise the NMI watchdog - if enabled - can
989 detect a hard lockup condition.
991 ==============================================================
995 Non-zero if the kernel has been tainted. Numeric values, which
996 can be ORed together:
998 1 - A module with a non-GPL license has been loaded, this
999 includes modules with no license.
1000 Set by modutils >= 2.4.9 and module-init-tools.
1001 2 - A module was force loaded by insmod -f.
1002 Set by modutils >= 2.4.9 and module-init-tools.
1003 4 - Unsafe SMP processors: SMP with CPUs not designed for SMP.
1004 8 - A module was forcibly unloaded from the system by rmmod -f.
1005 16 - A hardware machine check error occurred on the system.
1006 32 - A bad page was discovered on the system.
1007 64 - The user has asked that the system be marked "tainted". This
1008 could be because they are running software that directly modifies
1009 the hardware, or for other reasons.
1010 128 - The system has died.
1011 256 - The ACPI DSDT has been overridden with one supplied by the user
1012 instead of using the one provided by the hardware.
1013 512 - A kernel warning has occurred.
1014 1024 - A module from drivers/staging was loaded.
1015 2048 - The system is working around a severe firmware bug.
1016 4096 - An out-of-tree module has been loaded.
1017 8192 - An unsigned module has been loaded in a kernel supporting module
1019 16384 - A soft lockup has previously occurred on the system.
1020 32768 - The kernel has been live patched.
1022 ==============================================================
1026 This value controls the maximum number of threads that can be created
1029 During initialization the kernel sets this value such that even if the
1030 maximum number of threads is created, the thread structures occupy only
1031 a part (1/8th) of the available RAM pages.
1033 The minimum value that can be written to threads-max is 20.
1034 The maximum value that can be written to threads-max is given by the
1035 constant FUTEX_TID_MASK (0x3fffffff).
1036 If a value outside of this range is written to threads-max an error
1039 The value written is checked against the available RAM pages. If the
1040 thread structures would occupy too much (more than 1/8th) of the
1041 available RAM pages threads-max is reduced accordingly.
1043 ==============================================================
1045 unprivileged_bpf_disabled:
1047 Writing 1 to this entry will disable unprivileged calls to bpf();
1048 once disabled, calling bpf() without CAP_SYS_ADMIN will return
1049 -EPERM. Once set to 1, this can't be cleared from the running kernel
1052 Writing 2 to this entry will also disable unprivileged calls to bpf(),
1053 however, an admin can still change this setting later on, if needed, by
1054 writing 0 or 1 to this entry.
1056 If BPF_UNPRIV_DEFAULT_OFF is enabled in the kernel config, then this
1057 entry will default to 2 instead of 0.
1059 0 - Unprivileged calls to bpf() are enabled
1060 1 - Unprivileged calls to bpf() are disabled without recovery
1061 2 - Unprivileged calls to bpf() are disabled
1063 ==============================================================
1067 The value in this file affects behavior of handling NMI. When the
1068 value is non-zero, unknown NMI is trapped and then panic occurs. At
1069 that time, kernel debugging information is displayed on console.
1071 NMI switch that most IA32 servers have fires unknown NMI up, for
1072 example. If a system hangs up, try pressing the NMI switch.
1074 ==============================================================
1078 Number of kernel warnings after which the kernel should panic when
1079 ``panic_on_warn`` is not set. Setting this to 0 disables checking
1080 the warning count. Setting this to 1 has the same effect as setting
1081 ``panic_on_warn=1``. The default value is 0.
1083 ==============================================================
1087 This parameter can be used to disable or enable the soft lockup detector
1088 _and_ the NMI watchdog (i.e. the hard lockup detector) at the same time.
1090 0 - disable both lockup detectors
1091 1 - enable both lockup detectors
1093 The soft lockup detector and the NMI watchdog can also be disabled or
1094 enabled individually, using the soft_watchdog and nmi_watchdog parameters.
1095 If the watchdog parameter is read, for example by executing
1097 cat /proc/sys/kernel/watchdog
1099 the output of this command (0 or 1) shows the logical OR of soft_watchdog
1102 ==============================================================
1106 This value can be used to control on which cpus the watchdog may run.
1107 The default cpumask is all possible cores, but if NO_HZ_FULL is
1108 enabled in the kernel config, and cores are specified with the
1109 nohz_full= boot argument, those cores are excluded by default.
1110 Offline cores can be included in this mask, and if the core is later
1111 brought online, the watchdog will be started based on the mask value.
1113 Typically this value would only be touched in the nohz_full case
1114 to re-enable cores that by default were not running the watchdog,
1115 if a kernel lockup was suspected on those cores.
1117 The argument value is the standard cpulist format for cpumasks,
1118 so for example to enable the watchdog on cores 0, 2, 3, and 4 you
1121 echo 0,2-4 > /proc/sys/kernel/watchdog_cpumask
1123 ==============================================================
1127 This value can be used to control the frequency of hrtimer and NMI
1128 events and the soft and hard lockup thresholds. The default threshold
1131 The softlockup threshold is (2 * watchdog_thresh). Setting this
1132 tunable to zero will disable lockup detection altogether.
1134 ==============================================================