proc_pid_io.5: Dewafflify

[thirdparty/man-pages.git] / man5 / proc_sys_vm.5

.\" Copyright (C) 1994, 1995, Daniel Quinlan <quinlan@yggdrasil.com>
.\" Copyright (C) 2002-2008, 2017, Michael Kerrisk <mtk.manpages@gmail.com>
.\" Copyright (C) , Andries Brouwer <aeb@cwi.nl>
.\" Copyright (C) 2023, Alejandro Colomar <alx@kernel.org>
.\"
.\" SPDX-License-Identifier: GPL-3.0-or-later
.\"
.TH proc_sys_vm 5 (date) "Linux man-pages (unreleased)"
.SH NAME
/proc/sys/vm/ \- virtual memory subsystem
.SH DESCRIPTION
.TP
.I /proc/sys/vm/
This directory contains files for memory management tuning, buffer, and
cache management.
.TP
.IR /proc/sys/vm/admin_reserve_kbytes " (since Linux 3.10)"
.\" commit 4eeab4f5580d11bffedc697684b91b0bca0d5009
This file defines the amount of free memory (in KiB) on the system that
should be reserved for users with the capability
.BR CAP_SYS_ADMIN .
.IP
The default value in this file is the minimum of [3% of free pages, 8MiB]
expressed as KiB.
The default is intended to provide enough for the superuser
to log in and kill a process, if necessary,
under the default overcommit 'guess' mode (i.e., 0 in
.IR /proc/sys/vm/overcommit_memory ).
.IP
Systems running in "overcommit never" mode (i.e., 2 in
.IR /proc/sys/vm/overcommit_memory )
should increase the value in this file to account
for the full virtual memory size of the programs used to recover (e.g.,
.BR login (1)
.BR ssh (1),
and
.BR top (1))
Otherwise, the superuser may not be able to log in to recover the system.
For example, on x86-64 a suitable value is 131072 (128MiB reserved).
.IP
Changing the value in this file takes effect whenever
an application requests memory.
.TP
.IR /proc/sys/vm/compact_memory " (since Linux 2.6.35)"
When 1 is written to this file, all zones are compacted such that free
memory is available in contiguous blocks where possible.
The effect of this action can be seen by examining
.IR /proc/buddyinfo .
.IP
Present only if the kernel was configured with
.BR CONFIG_COMPACTION .
.TP
.IR /proc/sys/vm/drop_caches " (since Linux 2.6.16)"
Writing to this file causes the kernel to drop clean caches, dentries, and
inodes from memory, causing that memory to become free.
This can be useful for memory management testing and
performing reproducible filesystem benchmarks.
Because writing to this file causes the benefits of caching to be lost,
it can degrade overall system performance.
.IP
To free pagecache, use:
.IP
.in +4n
.EX
echo 1 > /proc/sys/vm/drop_caches
.EE
.in
.IP
To free dentries and inodes, use:
.IP
.in +4n
.EX
echo 2 > /proc/sys/vm/drop_caches
.EE
.in
.IP
To free pagecache, dentries, and inodes, use:
.IP
.in +4n
.EX
echo 3 > /proc/sys/vm/drop_caches
.EE
.in
.IP
Because writing to this file is a nondestructive operation and dirty objects
are not freeable, the
user should run
.BR sync (1)
first.
.TP
.IR  /proc/sys/vm/sysctl_hugetlb_shm_group " (since Linux 2.6.7)"
This writable file contains a group ID that is allowed
to allocate memory using huge pages.
If a process has a filesystem group ID or any supplementary group ID that
matches this group ID,
then it can make huge-page allocations without holding the
.B CAP_IPC_LOCK
capability; see
.BR memfd_create (2),
.BR mmap (2),
and
.BR shmget (2).
.TP
.IR /proc/sys/vm/legacy_va_layout " (since Linux 2.6.9)"
.\" The following is from Documentation/filesystems/proc.txt
If nonzero, this disables the new 32-bit memory-mapping layout;
the kernel will use the legacy (2.4) layout for all processes.
.TP
.IR /proc/sys/vm/memory_failure_early_kill " (since Linux 2.6.32)"
.\" The following is based on the text in Documentation/sysctl/vm.txt
Control how to kill processes when an uncorrected memory error
(typically a 2-bit error in a memory module)
that cannot be handled by the kernel
is detected in the background by hardware.
In some cases (like the page still having a valid copy on disk),
the kernel will handle the failure
transparently without affecting any applications.
But if there is no other up-to-date copy of the data,
it will kill processes to prevent any data corruptions from propagating.
.IP
The file has one of the following values:
.RS
.TP
.B 1
Kill all processes that have the corrupted-and-not-reloadable page mapped
as soon as the corruption is detected.
Note that this is not supported for a few types of pages,
such as kernel internally
allocated data or the swap cache, but works for the majority of user pages.
.TP
.B 0
Unmap the corrupted page from all processes and kill a process
only if it tries to access the page.
.RE
.IP
The kill is performed using a
.B SIGBUS
signal with
.I si_code
set to
.BR BUS_MCEERR_AO .
Processes can handle this if they want to; see
.BR sigaction (2)
for more details.
.IP
This feature is active only on architectures/platforms with advanced machine
check handling and depends on the hardware capabilities.
.IP
Applications can override the
.I memory_failure_early_kill
setting individually with the
.BR prctl (2)
.B PR_MCE_KILL
operation.
.IP
Present only if the kernel was configured with
.BR CONFIG_MEMORY_FAILURE .
.TP
.IR /proc/sys/vm/memory_failure_recovery " (since Linux 2.6.32)"
.\" The following is based on the text in Documentation/sysctl/vm.txt
Enable memory failure recovery (when supported by the platform).
.RS
.TP
.B 1
Attempt recovery.
.TP
.B 0
Always panic on a memory failure.
.RE
.IP
Present only if the kernel was configured with
.BR CONFIG_MEMORY_FAILURE .
.TP
.IR /proc/sys/vm/oom_dump_tasks " (since Linux 2.6.25)"
.\" The following is from Documentation/sysctl/vm.txt
Enables a system-wide task dump (excluding kernel threads) to be
produced when the kernel performs an OOM-killing.
The dump includes the following information
for each task (thread, process):
thread ID, real user ID, thread group ID (process ID),
virtual memory size, resident set size,
the CPU that the task is scheduled on,
oom_adj score (see the description of
.IR /proc/ pid /oom_adj ),
and command name.
This is helpful to determine why the OOM-killer was invoked
and to identify the rogue task that caused it.
.IP
If this contains the value zero, this information is suppressed.
On very large systems with thousands of tasks,
it may not be feasible to dump the memory state information for each one.
Such systems should not be forced to incur a performance penalty in
OOM situations when the information may not be desired.
.IP
If this is set to nonzero, this information is shown whenever the
OOM-killer actually kills a memory-hogging task.
.IP
The default value is 0.
.TP
.IR /proc/sys/vm/oom_kill_allocating_task " (since Linux 2.6.24)"
.\" The following is from Documentation/sysctl/vm.txt
This enables or disables killing the OOM-triggering task in
out-of-memory situations.
.IP
If this is set to zero, the OOM-killer will scan through the entire
tasklist and select a task based on heuristics to kill.
This normally selects a rogue memory-hogging task that
frees up a large amount of memory when killed.
.IP
If this is set to nonzero, the OOM-killer simply kills the task that
triggered the out-of-memory condition.
This avoids a possibly expensive tasklist scan.
.IP
If
.I /proc/sys/vm/panic_on_oom
is nonzero, it takes precedence over whatever value is used in
.IR /proc/sys/vm/oom_kill_allocating_task .
.IP
The default value is 0.
.TP
.IR /proc/sys/vm/overcommit_kbytes " (since Linux 3.14)"
.\" commit 49f0ce5f92321cdcf741e35f385669a421013cb7
This writable file provides an alternative to
.I /proc/sys/vm/overcommit_ratio
for controlling the
.I CommitLimit
when
.I /proc/sys/vm/overcommit_memory
has the value 2.
It allows the amount of memory overcommitting to be specified as
an absolute value (in kB),
rather than as a percentage, as is done with
.IR overcommit_ratio .
This allows for finer-grained control of
.I CommitLimit
on systems with extremely large memory sizes.
.IP
Only one of
.I overcommit_kbytes
or
.I overcommit_ratio
can have an effect:
if
.I overcommit_kbytes
has a nonzero value, then it is used to calculate
.IR CommitLimit ,
otherwise
.I overcommit_ratio
is used.
Writing a value to either of these files causes the
value in the other file to be set to zero.
.TP
.I /proc/sys/vm/overcommit_memory
This file contains the kernel virtual memory accounting mode.
Values are:
.RS
.IP
0: heuristic overcommit (this is the default)
.br
1: always overcommit, never check
.br
2: always check, never overcommit
.RE
.IP
In mode 0, calls of
.BR mmap (2)
with
.B MAP_NORESERVE
are not checked, and the default check is very weak,
leading to the risk of getting a process "OOM-killed".
.IP
In mode 1, the kernel pretends there is always enough memory,
until memory actually runs out.
One use case for this mode is scientific computing applications
that employ large sparse arrays.
Before Linux 2.6.0, any nonzero value implies mode 1.
.IP
In mode 2 (available since Linux 2.6), the total virtual address space
that can be allocated
.RI ( CommitLimit
in
.IR /proc/meminfo )
is calculated as
.IP
.in +4n
.EX
CommitLimit = (total_RAM \- total_huge_TLB) *
	      overcommit_ratio / 100 + total_swap
.EE
.in
.IP
where:
.RS
.IP \[bu] 3
.I total_RAM
is the total amount of RAM on the system;
.IP \[bu]
.I total_huge_TLB
is the amount of memory set aside for huge pages;
.IP \[bu]
.I overcommit_ratio
is the value in
.IR /proc/sys/vm/overcommit_ratio ;
and
.IP \[bu]
.I total_swap
is the amount of swap space.
.RE
.IP
For example, on a system with 16 GB of physical RAM, 16 GB
of swap, no space dedicated to huge pages, and an
.I overcommit_ratio
of 50, this formula yields a
.I CommitLimit
of 24 GB.
.IP
Since Linux 3.14, if the value in
.I /proc/sys/vm/overcommit_kbytes
is nonzero, then
.I CommitLimit
is instead calculated as:
.IP
.in +4n
.EX
CommitLimit = overcommit_kbytes + total_swap
.EE
.in
.IP
See also the description of
.I /proc/sys/vm/admin_reserve_kbytes
and
.IR /proc/sys/vm/user_reserve_kbytes .
.TP
.IR /proc/sys/vm/overcommit_ratio " (since Linux 2.6.0)"
This writable file defines a percentage by which memory
can be overcommitted.
The default value in the file is 50.
See the description of
.IR /proc/sys/vm/overcommit_memory .
.TP
.IR /proc/sys/vm/panic_on_oom " (since Linux 2.6.18)"
.\" The following is adapted from Documentation/sysctl/vm.txt
This enables or disables a kernel panic in
an out-of-memory situation.
.IP
If this file is set to the value 0,
the kernel's OOM-killer will kill some rogue process.
Usually, the OOM-killer is able to kill a rogue process and the
system will survive.
.IP
If this file is set to the value 1,
then the kernel normally panics when out-of-memory happens.
However, if a process limits allocations to certain nodes
using memory policies
.RB ( mbind (2)
.BR MPOL_BIND )
or cpusets
.RB ( cpuset (7))
and those nodes reach memory exhaustion status,
one process may be killed by the OOM-killer.
No panic occurs in this case:
because other nodes' memory may be free,
this means the system as a whole may not have reached
an out-of-memory situation yet.
.IP
If this file is set to the value 2,
the kernel always panics when an out-of-memory condition occurs.
.IP
The default value is 0.
1 and 2 are for failover of clustering.
Select either according to your policy of failover.
.TP
.I /proc/sys/vm/swappiness
.\" The following is from Documentation/sysctl/vm.txt
The value in this file controls how aggressively the kernel will swap
memory pages.
Higher values increase aggressiveness, lower values
decrease aggressiveness.
The default value is 60.
.TP
.IR /proc/sys/vm/user_reserve_kbytes " (since Linux 3.10)"
.\" commit c9b1d0981fcce3d9976d7b7a56e4e0503bc610dd
Specifies an amount of memory (in KiB) to reserve for user processes.
This is intended to prevent a user from starting a single memory hogging
process, such that they cannot recover (kill the hog).
The value in this file has an effect only when
.I /proc/sys/vm/overcommit_memory
is set to 2 ("overcommit never" mode).
In this case, the system reserves an amount of memory that is the minimum
of [3% of current process size,
.IR user_reserve_kbytes ].
.IP
The default value in this file is the minimum of [3% of free pages, 128MiB]
expressed as KiB.
.IP
If the value in this file is set to zero,
then a user will be allowed to allocate all free memory with a single process
(minus the amount reserved by
.IR /proc/sys/vm/admin_reserve_kbytes ).
Any subsequent attempts to execute a command will result in
"fork: Cannot allocate memory".
.IP
Changing the value in this file takes effect whenever
an application requests memory.
.TP
.IR /proc/sys/vm/unprivileged_userfaultfd " (since Linux 5.2)"
.\" cefdca0a86be517bc390fc4541e3674b8e7803b0
This (writable) file exposes a flag that controls whether
unprivileged processes are allowed to employ
.BR userfaultfd (2).
If this file has the value 1, then unprivileged processes may use
.BR userfaultfd (2).
If this file has the value 0, then only processes that have the
.B CAP_SYS_PTRACE
capability may employ
.BR userfaultfd (2).
The default value in this file is 1.
.SH SEE ALSO
.BR proc (5),
.BR proc_sys (5)