1 .\" Copyright (C) 1994, 1995 by Daniel Quinlan (quinlan@yggdrasil.com)
2 .\" and Copyright (C) 2002-2008 Michael Kerrisk <mtk.manpages@gmail.com>
3 .\" with networking additions from Alan Cox (A.Cox@swansea.ac.uk)
4 .\" and scsi additions from Michael Neuffer (neuffer@mail.uni-mainz.de)
5 .\" and sysctl additions from Andries Brouwer (aeb@cwi.nl)
6 .\" and System V IPC (as well as various other) additions from
7 .\" Michael Kerrisk <mtk.manpages@gmail.com>
9 .\" %%%LICENSE_START(GPLv2+_DOC_FULL)
10 .\" This is free documentation; you can redistribute it and/or
11 .\" modify it under the terms of the GNU General Public License as
12 .\" published by the Free Software Foundation; either version 2 of
13 .\" the License, or (at your option) any later version.
15 .\" The GNU General Public License's references to "object code"
16 .\" and "executables" are to be interpreted as the output of any
17 .\" document formatting or typesetting system, including
18 .\" intermediate and printed output.
20 .\" This manual is distributed in the hope that it will be useful,
21 .\" but WITHOUT ANY WARRANTY; without even the implied warranty of
22 .\" MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
23 .\" GNU General Public License for more details.
25 .\" You should have received a copy of the GNU General Public
26 .\" License along with this manual; if not, see
27 .\" <http://www.gnu.org/licenses/>.
30 .\" Modified 1995-05-17 by faith@cs.unc.edu
31 .\" Minor changes by aeb and Marty Leisner (leisner@sdsp.mc.xerox.com).
32 .\" Modified 1996-04-13, 1996-07-22 by aeb@cwi.nl
33 .\" Modified 2001-12-16 by rwhron@earthlink.net
34 .\" Modified 2002-07-13 by jbelton@shaw.ca
35 .\" Modified 2002-07-22, 2003-05-27, 2004-04-06, 2004-05-25
36 .\" by Michael Kerrisk <mtk.manpages@gmail.com>
37 .\" 2004-11-17, mtk -- updated notes on /proc/loadavg
38 .\" 2004-12-01, mtk, rtsig-max and rtsig-nr went away in 2.6.8
39 .\" 2004-12-14, mtk, updated 'statm', and fixed error in order of list
40 .\" 2005-05-12, mtk, updated 'stat'
41 .\" 2005-07-13, mtk, added /proc/sys/fs/mqueue/*
42 .\" 2005-09-16, mtk, Added /proc/sys/fs/suid_dumpable
43 .\" 2005-09-19, mtk, added /proc/zoneinfo
44 .\" 2005-03-01, mtk, moved /proc/sys/fs/mqueue/* material to mq_overview.7.
45 .\" 2008-06-05, mtk, Added /proc/[pid]/oom_score, /proc/[pid]/oom_adj,
46 .\" /proc/[pid]/limits, /proc/[pid]/mountinfo, /proc/[pid]/mountstats,
47 .\" and /proc/[pid]/fdinfo/*.
48 .\" 2008-06-19, mtk, Documented /proc/[pid]/status.
49 .\" 2008-07-15, mtk, added /proc/config.gz
51 .\" FIXME 2.6.13 seems to have /proc/vmcore implemented
52 .\" in the source code, but there is no option available under
53 .\" 'make xconfig'; eventually this should be fixed, and then info
54 .\" from the patch-2.6.13 and change log could be used to write an
55 .\" entry in this man page.
56 .\" Needs CONFIG_VMCORE
58 .\" FIXME cross check against Documentation/filesystems/proc.txt
59 .\" to see what information could be imported from that file
62 .TH PROC 5 2014-04-12 "Linux" "Linux Programmer's Manual"
64 proc \- process information pseudo-filesystem
68 filesystem is a pseudo-filesystem which provides an interface to
69 kernel data structures.
70 It is commonly mounted at
72 Most of it is read-only, but some files allow kernel variables to be
75 The following list describes many of the files and directories under the
81 There is a numerical subdirectory for each running process; the
82 subdirectory is named by the process ID.
83 Each such subdirectory contains the following
84 pseudo-files and directories.
85 .\" FIXME Describe /proc/[pid]/attr and
86 .\" /proc/[pid]/task/[tid]/attr
87 .\" This is a directory
92 .IR /proc/[pid]/auxv " (since 2.6.0-test7)"
93 This contains the contents of the ELF interpreter information passed
94 to the process at exec time.
95 The format is one \fIunsigned long\fP ID
96 plus one \fIunsigned long\fP value for each entry.
97 The last entry contains two zeros.
99 .IR /proc/[pid]/cgroup " (since Linux 2.6.24)"
100 .\" Info in Documentation/cgroups/cgroups.txt
101 This file describes control groups to which the process/task belongs.
102 For each cgroup hierarchy there is one entry containing
103 colon-separated fields of the form:
107 5:cpuacct,cpu,cpuset:/daemons
111 The colon-separated fields are, from left to right:
116 set of subsystems bound to the hierarchy
118 control group in the hierarchy to which the process belongs
121 This file is present only if the
123 kernel configuration option is enabled.
125 .IR /proc/[pid]/clear_refs " (since Linux 2.6.22)"
126 .\" commit b813e931b4c8235bb42e301096ea97dbdee3e8fe (2.6.22)
127 .\" commit 398499d5f3613c47f2143b8c54a04efb5d7a6da9 (2.6.32)
128 .\" commit 040fa02077de01c7e08fa75be6125e4ca5636011 (3.11)
130 .\" "Clears page referenced bits shown in smaps output"
131 .\" write-only, writable only by the owner of the process
133 This is a write-only file, writable only by owner of the process.
135 The following values may be written to the file:
138 1 (since Linux 2.6.22)
139 .\" Internally: CLEAR_REFS_ALL
140 Reset the PG_Referenced and ACCESSED/YOUNG
141 bits for all the pages associated with the process.
142 (Before kernel 2.6.32, writing any nonzero value to this file
145 2 (since Linux 2.6.32)
146 .\" Internally: CLEAR_REFS_ANON
147 Reset the PG_Referenced and ACCESSED/YOUNG
148 bits for all anonymous pages associated with the process.
150 3 (since Linux 2.6.32)
151 .\" Internally: CLEAR_REFS_MAPPED
152 Reset the PG_Referenced and ACCESSED/YOUNG
153 bits for all file-mapped pages associated with the process.
156 Clearing the PG_Referenced and ACCESSED/YOUNG bits provides a method
157 to measure approximately how much memory a process is using.
158 One first inspects the values in the "Referenced:" fields
159 for the VMAs shown in
160 .IR /proc/[pid]/smaps
161 to get an idea of the memory footprint of the
163 One then clears the PG_Referenced and ACCESSED/YOUNG bits
164 and, after some measured time interval,
165 once again inspects the values in the "Referenced:" fields
166 to get an idea of the change in memory footprint of the
167 process during the measured interval.
168 If one is interested only in inspecting the selected mapping types,
169 then the value 2 or 3 can be used instead of 1.
171 A further value can be written to affect a different bit:
175 Clear the soft-dirty bit for all the pages associated with the process.
176 .\" Internally: CLEAR_REFS_SOFT_DIRTY
177 This is used (in conjunction with
178 .IR /proc/[pid]/pagemap )
179 by the check-point restore system to discover which pages of a process
180 have been dirtied since the file
181 .IR /proc/[pid]/clear_refs
186 .IR /proc/[pid]/clear_refs
187 other than those listed above has no effect.
190 .IR /proc/[pid]/clear_refs
191 file is present only if the
192 .B CONFIG_PROC_PAGE_MONITOR
193 kernel configuration option is enabled.
195 .I /proc/[pid]/cmdline
196 This holds the complete command line for the process,
197 unless the process is a zombie.
198 .\" In 2.3.26, this also used to be true if the process was swapped out.
199 In the latter case, there is nothing in this file:
200 that is, a read on this file will return 0 characters.
201 The command-line arguments appear in this file as a set of
202 strings separated by null bytes (\(aq\\0\(aq),
203 with a further null byte after the last string.
205 .IR /proc/[pid]/coredump_filter " (since kernel 2.6.23)"
209 .IR /proc/[pid]/cpuset " (since kernel 2.6.12)"
210 .\" and/proc/[pid]/task/[tid]/cpuset
215 This is a symbolic link to the current working directory of the process.
216 To find out the current working directory of process 20,
217 for instance, you can do this:
221 .RB "$" " cd /proc/20/cwd; /bin/pwd"
227 command is often a shell built-in, and might
234 .\" The following was still true as at kernel 2.6.13
235 In a multithreaded process, the contents of this symbolic link
236 are not available if the main thread has already terminated
237 (typically by calling
238 .BR pthread_exit (3)).
240 .I /proc/[pid]/environ
241 This file contains the environment for the process.
242 The entries are separated by null bytes (\(aq\\0\(aq),
243 and there may be a null byte at the end.
244 Thus, to print out the environment of process 1, you would do:
249 .RB "$" " strings /proc/1/environ"
255 Under Linux 2.2 and later, this file is a symbolic link
256 containing the actual pathname of the executed command.
257 This symbolic link can be dereferenced normally; attempting to open
258 it will open the executable.
261 to run another copy of the same executable as is being run by
263 .\" The following was still true as at kernel 2.6.13
264 In a multithreaded process, the contents of this symbolic link
265 are not available if the main thread has already terminated
266 (typically by calling
267 .BR pthread_exit (3)).
269 Under Linux 2.0 and earlier
271 is a pointer to the binary which was executed,
272 and appears as a symbolic link.
275 call on this file under Linux 2.0 returns a string in the format:
279 For example, [0301]:1502 would be inode 1502 on device major 03 (IDE,
280 MFM, etc. drives) minor 01 (first partition on the first drive).
285 option can be used to locate the file.
288 This is a subdirectory containing one entry for each file which the
289 process has open, named by its file descriptor, and which is a
290 symbolic link to the actual file.
291 Thus, 0 is standard input, 1 standard output, 2 standard error, and so on.
293 For file descriptors for pipes and sockets,
294 the entries will be symbolic links whose content is the
295 file type with the inode.
298 call on this file returns a string in the format:
304 will be a socket and its inode is 2248868.
305 For sockets, that inode can be used to find more information
306 in one of the files under
309 For file descriptors that have no corresponding inode
310 (e.g., file descriptors produced by
311 .BR epoll_create (2),
313 .BR inotify_init (2),
317 the entry will be a symbolic link with contents of the form
319 anon_inode:<file-type>
323 is surrounded by square brackets.
325 For example, an epoll file descriptor will have a symbolic link
326 whose content is the string
327 .IR "anon_inode:[eventpoll]" .
329 .\"The following was still true as at kernel 2.6.13
330 In a multithreaded process, the contents of this directory
331 are not available if the main thread has already terminated
332 (typically by calling
333 .BR pthread_exit (3)).
335 Programs that will take a filename as a command-line argument,
336 but will not take input from standard input if no argument is supplied,
337 or that write to a file named as a command-line argument,
338 but will not send their output to standard output
339 if no argument is supplied, can nevertheless be made to use
340 standard input or standard out using
342 For example, assuming that
344 is the flag designating an input file and
346 is the flag designating an output file:
350 .RB "$" " foobar \-i /proc/self/fd/0 \-o /proc/self/fd/1 ..."
354 and you have a working filter.
355 .\" The following is not true in my tests (MTK):
356 .\" Note that this will not work for
357 .\" programs that seek on their files, as the files in the fd directory
358 .\" are not seekable.
361 is approximately the same as
363 in some UNIX and UNIX-like systems.
364 Most Linux MAKEDEV scripts symbolically link
370 Most systems provide symbolic links
375 which respectively link to the files
382 Thus the example command above could be written as:
386 .RB "$" " foobar \-i /dev/stdin \-o /dev/stdout ..."
389 .\" FIXME Describe /proc/[pid]/loginuid
390 .\" Added in 2.6.11; updating requires CAP_AUDIT_CONTROL
391 .\" CONFIG_AUDITSYSCALL
393 .IR /proc/[pid]/fdinfo/ " (since kernel 2.6.22)"
394 This is a subdirectory containing one entry for each file which the
395 process has open, named by its file descriptor.
396 The contents of each file can be read to obtain information
397 about the corresponding file descriptor, for example:
401 .RB "$" " cat /proc/12015/fdinfo/4"
409 field is a decimal number showing the current file offset.
412 field is an octal number that displays the
413 file access mode and file status flags (see
416 The files in this directory are readable only by the owner of the process.
418 .\" Certain file types include additional info; see
419 .\" Documentation/filesystems/proc.txt
421 .\" Especially interesting is this:
423 .\" commit ab49bdecc3ebb46ab661f5f05d5c5ea9606406c6
424 .\" Author: Cyrill Gorcunov <gorcunov@openvz.org>
425 .\" Date: Mon Dec 17 16:05:06 2012 -0800
427 .\" Basically, the /proc/PID/fdinfo/ entry for an inotify FD
428 .\" includes the file handles for all watched FDs
431 .IR /proc/[pid]/io " (since kernel 2.6.20)"
432 .\" commit 7c3ab7381e79dfc7db14a67c6f4f3285664e1ec2
433 This file contains I/O statistics for the process, for example:
437 .RB "#" " cat /proc/3828/io"
443 write_bytes: 323932160
444 cancelled_write_bytes: 0
448 The fields are as follows:
451 .IR rchar ": characters read"
452 The number of bytes which this task has caused to be read from storage.
453 This is simply the sum of bytes which this process passed to
455 and similar system calls.
456 It includes things such as terminal I/O and
457 is unaffected by whether or not actual
458 physical disk I/O was required (the read might have been satisfied from
461 .IR wchar ": characters written"
462 The number of bytes which this task has caused, or shall cause to be written
464 Similar caveats apply here as with
467 .IR syscr ": read syscalls"
468 Attempt to count the number of read I/O operations\(emthat is,
474 .IR syscw ": write syscalls"
475 Attempt to count the number of write I/O operations\(emthat is,
481 .IR read_bytes ": bytes read"
482 Attempt to count the number of bytes which this process really did cause to
483 be fetched from the storage layer.
484 This is accurate for block-backed filesystems.
486 .IR write_bytes ": bytes written"
487 Attempt to count the number of bytes which this process caused to be sent to
490 .IR cancelled_write_bytes :
491 The big inaccuracy here is truncate.
492 If a process writes 1MB to a file and then deletes the file,
493 it will in fact perform no writeout.
494 But it will have been accounted as having caused 1MB of write.
495 In other words: this field represents the number of bytes which this process
496 caused to not happen, by truncating pagecache.
497 A task can cause "negative" I/O too.
498 If this task truncates some dirty pagecache,
499 some I/O which another task has been accounted for
502 will not be happening.
506 In the current implementation, things are a bit racy on 32-bit systems:
507 if process A reads process B's
509 while process B is updating one of these 64-bit counters,
510 process A could see an intermediate result.
512 .IR /proc/[pid]/limits " (since kernel 2.6.24)"
513 This file displays the soft limit, hard limit, and units of measurement
514 for each of the process's resource limits (see
516 Up to and including Linux 2.6.35,
517 this file is protected to allow reading only by the real UID of the process.
519 .\" commit 3036e7b490bf7878c6dae952eec5fb87b1106589
520 this file is readable by all users on the system.
522 .IR /proc/[pid]/map_files/ " (since kernel 3.3)
523 .\" commit 640708a2cff7f81e246243b0073c66e6ece7e53e
524 This subdirectory contains entries corresponding to memory-mapped
527 Entries are named by memory region start and end
528 address pair (expressed as hexadecimal numbers),
529 and are symbolic links to the mapped files themselves.
530 Here is an example, with the output wrapped and reformatted to fit on an 80-column display:
534 .RB "$" " ls -l /proc/self/map_files/"
535 lr\-\-\-\-\-\-\-\-. 1 root root 64 Apr 16 21:31
536 3252e00000\-3252e20000 \-> /usr/lib64/ld-2.15.so
541 Although these entries are present for memory regions that were
544 flag, the way anonymous shared memory (regions created with the
545 .B MAP_ANON | MAP_SHARED
547 is implemented in Linux
548 means that such regions also appear on this directory.
549 Here is an example where the target file is the deleted
556 lrw\-\-\-\-\-\-\-. 1 root root 64 Apr 16 21:33
557 7fc075d2f000\-7fc075e6f000 \-> /dev/zero (deleted)
561 This directory appears only if the
562 .B CONFIG_CHECKPOINT_RESTORE
563 kernel configuration option is enabled.
566 A file containing the currently mapped memory regions and their access
570 for some further information about memory mappings.
572 The format of the file is:
578 .I "address perms offset dev inode pathname"
579 00400000-00452000 r-xp 00000000 08:02 173521 /usr/bin/dbus-daemon
580 00651000-00652000 r--p 00051000 08:02 173521 /usr/bin/dbus-daemon
581 00652000-00655000 rw-p 00052000 08:02 173521 /usr/bin/dbus-daemon
582 00e03000-00e24000 rw-p 00000000 00:00 0 [heap]
583 00e24000-011f7000 rw-p 00000000 00:00 0 [heap]
585 35b1800000-35b1820000 r-xp 00000000 08:02 135522 /usr/lib64/ld-2.15.so
586 35b1a1f000-35b1a20000 r--p 0001f000 08:02 135522 /usr/lib64/ld-2.15.so
587 35b1a20000-35b1a21000 rw-p 00020000 08:02 135522 /usr/lib64/ld-2.15.so
588 35b1a21000-35b1a22000 rw-p 00000000 00:00 0
589 35b1c00000-35b1dac000 r-xp 00000000 08:02 135870 /usr/lib64/libc-2.15.so
590 35b1dac000-35b1fac000 ---p 001ac000 08:02 135870 /usr/lib64/libc-2.15.so
591 35b1fac000-35b1fb0000 r--p 001ac000 08:02 135870 /usr/lib64/libc-2.15.so
592 35b1fb0000-35b1fb2000 rw-p 001b0000 08:02 135870 /usr/lib64/libc-2.15.so
594 f2c6ff8c000-7f2c7078c000 rw-p 00000000 00:00 0 [stack:986]
596 7fffb2c0d000-7fffb2c2e000 rw-p 00000000 00:00 0 [stack]
597 7fffb2d48000-7fffb2d49000 r-xp 00000000 00:00 0 [vdso]
603 field is the address space in the process that the mapping occupies.
606 field is a set of permissions:
614 p = private (copy on write)
620 field is the offset into the file/whatever;
625 is the inode on that device.
626 0 indicates that no inode is associated with the memory region,
627 as would be the case with BSS (uninitialized data).
631 field will usually be the file that is backing the mapping.
633 you can easily coordinate with the
635 field by looking at the
636 Offset field in the ELF program headers
637 .RI ( "readelf\ \-l" ).
639 There are additional helpful pseudo-paths:
643 The initial process's (also known as the main thread's) stack.
645 .IR [stack:<tid>] " (since Linux 3.4)"
646 .\" commit b76437579d1344b612cf1851ae610c636cec7db0
647 A thread's stack (where the
650 It corresponds to the
651 .IR /proc/[pid]/task/[tid]/
655 The virtual dynamically linked shared object.
665 this is an anonymous mapping as obtained via the
668 There is no easy way to coordinate this back to a process's source,
669 short of running it through
674 Under Linux 2.0 there is no field giving pathname.
677 This file can be used to access the pages of a process's memory through
683 .IR /proc/[pid]/mountinfo " (since Linux 2.6.26)"
684 .\" This info adapted from Documentation/filesystems/proc.txt
685 This file contains information about mount points.
686 It contains lines of the form:
690 36 35 98:0 /mnt1 /mnt2 rw,noatime master:1 - ext3 /dev/root rw,errors=continue
691 (1)(2)(3) (4) (5) (6) (7) (8) (9) (10) (11)
695 The numbers in parentheses are labels for the descriptions below:
699 mount ID: unique identifier of the mount (may be reused after
703 parent ID: ID of parent mount (or of self for the top of the mount tree).
706 major:minor: value of
708 for files on filesystem (see
712 root: root of the mount within the filesystem.
715 mount point: mount point relative to the process's root.
718 mount options: per-mount options.
721 optional fields: zero or more fields of the form "tag[:value]".
724 separator: marks the end of the optional fields.
727 filesystem type: name of filesystem in the form "type[.subtype]".
730 mount source: filesystem-specific information or "none".
733 super options: per-superblock options.
736 Parsers should ignore all unrecognized optional fields.
737 Currently the possible optional fields are:
741 mount is shared in peer group X
744 mount is slave to peer group X
747 mount is slave and receives propagation from peer group X (*)
753 (*) X is the closest dominant peer group under the process's root.
754 If X is the immediate master of the mount,
755 or if there is no dominant peer group under the same root,
756 then only the "master:X" field is present
757 and not the "propagate_from:X" field.
759 For more information on mount propagation see:
760 .I Documentation/filesystems/sharedsubtree.txt
761 in the Linux kernel source tree.
763 .IR /proc/[pid]/mounts " (since Linux 2.4.19)"
764 This is a list of all the filesystems currently mounted in the
765 process's mount namespace.
766 The format of this file is documented in
768 Since kernel version 2.6.15, this file is pollable:
769 after opening the file for reading, a change in this file
770 (i.e., a filesystem mount or unmount) causes
772 to mark the file descriptor as readable, and
776 mark the file as having an error condition.
778 .IR /proc/[pid]/mountstats " (since Linux 2.6.17)"
779 This file exports information (statistics, configuration information)
780 about the mount points in the process's name space.
781 Lines in this file have the form:
784 device /dev/sda7 mounted on /home with fstype ext3 [statistics]
788 The fields in each line are:
792 The name of the mounted device
793 (or "nodevice" if there is no corresponding device).
796 The mount point within the filesystem tree.
802 Optional statistics and configuration information.
803 Currently (as at Linux 2.6.26), only NFS filesystems export
804 information via this field.
807 This file is readable only by the owner of the process.
809 .IR /proc/[pid]/ns/ " (since Linux 3.0)"
810 .\" See commit 6b4e306aa3dc94a0545eb9279475b1ab6209a31f
811 This is a subdirectory containing one entry for each namespace that
812 supports being manipulated by
814 For information about namespaces, see
817 .IR /proc/[pid]/ns/ipc " (since Linux 3.0)"
818 Bind mounting this file (see
820 to somewhere else in the filesystem keeps
821 the IPC namespace of the process specified by
823 alive even if all processes currently in the namespace terminate.
825 Opening this file returns a file handle for the IPC namespace
826 of the process specified by
828 As long as this file descriptor remains open,
829 the IPC namespace will remain alive,
830 even if all processes in the namespace terminate.
831 The file descriptor can be passed to
834 .IR /proc/[pid]/ns/net " (since Linux 3.0)"
835 Bind mounting this file (see
837 to somewhere else in the filesystem keeps
838 the network namespace of the process specified by
840 alive even if all processes in the namespace terminate.
842 Opening this file returns a file handle for the network namespace
843 of the process specified by
845 As long as this file descriptor remains open,
846 the network namespace will remain alive,
847 even if all processes in the namespace terminate.
848 The file descriptor can be passed to
851 .IR /proc/[pid]/ns/uts " (since Linux 3.0)"
852 Bind mounting this file (see
854 to somewhere else in the filesystem keeps
855 the UTS namespace of the process specified by
857 alive even if all processes currently in the namespace terminate.
859 Opening this file returns a file handle for the UTS namespace
860 of the process specified by
862 As long as this file descriptor remains open,
863 the UTS namespace will remain alive,
864 even if all processes in the namespace terminate.
865 The file descriptor can be passed to
868 .IR /proc/[pid]/numa_maps " (since Linux 2.6.14)"
872 .IR /proc/[pid]/oom_adj " (since Linux 2.6.11)"
873 This file can be used to adjust the score used to select which process
874 should be killed in an out-of-memory (OOM) situation.
875 The kernel uses this value for a bit-shift operation of the process's
878 valid values are in the range \-16 to +15,
879 plus the special value \-17,
880 which disables OOM-killing altogether for this process.
881 A positive score increases the likelihood of this
882 process being killed by the OOM-killer;
883 a negative score decreases the likelihood.
885 The default value for this file is 0;
886 a new process inherits its parent's
889 A process must be privileged
890 .RB ( CAP_SYS_RESOURCE )
893 Since Linux 2.6.36, use of this file is deprecated in favor of
894 .IR /proc/[pid]/oom_score_adj .
896 .IR /proc/[pid]/oom_score " (since Linux 2.6.11)"
897 .\" See mm/oom_kill.c::badness() in the 2.6.25 sources
898 This file displays the current score that the kernel gives to
899 this process for the purpose of selecting a process
901 A higher score means that the process is more likely to be
902 selected by the OOM-killer.
903 The basis for this score is the amount of memory used by the process,
904 with increases (+) or decreases (\-) for factors including:
905 .\" See mm/oom_kill.c::badness() in the 2.6.25 sources
908 whether the process creates a lot of children using
912 whether the process has been running a long time,
913 or has used a lot of CPU time (\-);
915 whether the process has a low nice value (i.e., > 0) (+);
917 whether the process is privileged (\-); and
918 .\" More precisely, if it has CAP_SYS_ADMIN or CAP_SYS_RESOURCE
920 whether the process is making direct hardware access (\-).
921 .\" More precisely, if it has CAP_SYS_RAWIO
926 also reflects the adjustment specified by the
930 setting for the process.
932 .IR /proc/[pid]/oom_score_adj " (since Linux 2.6.36)"
933 .\" Text taken from 3.7 Documentation/filesystems/proc.txt
934 This file can be used to adjust the badness heuristic used to select which
935 process gets killed in out-of-memory conditions.
937 The badness heuristic assigns a value to each candidate task ranging from 0
938 (never kill) to 1000 (always kill) to determine which process is targeted.
939 The units are roughly a proportion along that range of
940 allowed memory the process may allocate from,
941 based on an estimation of its current memory and swap use.
942 For example, if a task is using all allowed memory,
943 its badness score will be 1000.
944 If it is using half of its allowed memory, its score will be 500.
946 There is an additional factor included in the badness score: root
947 processes are given 3% extra memory over other tasks.
949 The amount of "allowed" memory depends on the context
950 in which the OOM-killer was called.
951 If it is due to the memory assigned to the allocating task's cpuset
953 the allowed memory represents the set of mems assigned to that
956 If it is due to a mempolicy's node(s) being exhausted,
957 the allowed memory represents the set of mempolicy nodes.
958 If it is due to a memory limit (or swap limit) being reached,
959 the allowed memory is that configured limit.
960 Finally, if it is due to the entire system being out of memory, the
961 allowed memory represents all allocatable resources.
965 is added to the badness score before it
966 is used to determine which task to kill.
967 Acceptable values range from \-1000
968 (OOM_SCORE_ADJ_MIN) to +1000 (OOM_SCORE_ADJ_MAX).
969 This allows user space to control the preference for OOM-killing,
970 ranging from always preferring a certain
971 task or completely disabling it from OOM-killing.
972 The lowest possible value, \-1000, is
973 equivalent to disabling OOM-killing entirely for that task,
974 since it will always report a badness score of 0.
976 Consequently, it is very simple for user space to define
977 the amount of memory to consider for each task.
980 value of +500, for example,
981 is roughly equivalent to allowing the remainder of tasks sharing the
982 same system, cpuset, mempolicy, or memory controller resources
983 to use at least 50% more memory.
984 A value of \-500, on the other hand, would be roughly
985 equivalent to discounting 50% of the task's
986 allowed memory from being considered as scoring against the task.
988 For backward compatibility with previous kernels,
989 .I /proc/[pid]/oom_adj
990 can still be used to tune the badness score.
996 .IR /proc/[pid]/oom_score_adj
998 .IR /proc/[pid]/oom_adj
999 will change the other with its scaled value.
1000 .\" FIXME Describe /proc/[pid]/pagemap
1002 .\" CONFIG_PROC_PAGE_MONITOR
1005 UNIX and Linux support the idea of a per-process root of the
1006 filesystem, set by the
1009 This file is a symbolic link that points to the process's
1010 root directory, and behaves in the same way as
1015 .\" The following was still true as at kernel 2.6.13
1016 In a multithreaded process, the contents of this symbolic link
1017 are not available if the main thread has already terminated
1018 (typically by calling
1019 .BR pthread_exit (3)).
1020 .\" FIXME Describe /proc/[pid]/seccomp
1022 .\" FIXME Describe /proc/[pid]/sessionid
1023 .\" Added in 2.6.25; read-only; only readable by real UID
1024 .\" CONFIG_AUDITSYSCALL
1025 .\" FIXME Describe /proc/[pid]/sched
1027 .\" CONFIG_SCHED_DEBUG, and additional fields if CONFIG_SCHEDSTATS
1028 .\" Displays various scheduling parameters
1029 .\" This file can be written, to reset stats
1030 .\" FIXME Describe /proc/[pid]/schedstats and
1031 .\" /proc/[pid]/task/[tid]/schedstats
1033 .\" CONFIG_SCHEDSTATS
1035 .IR /proc/[pid]/smaps " (since Linux 2.6.14)"
1036 This file shows memory consumption for each of the process's mappings.
1037 For each mapping there is a series of lines such as the following:
1041 00400000-0048a000 r-xp 00000000 fd:03 960637 /bin/bash
1045 Shared_Clean: 452 kB
1053 KernelPageSize: 4 kB
1059 The first of these lines shows the same information as is displayed
1061 .IR /proc/[pid]/maps .
1062 The remaining lines show the size of the mapping,
1063 the amount of the mapping that is currently resident in RAM ("Rss"),
1064 the process' proportional share of this mapping ("Pss"),
1065 the number of clean and dirty shared pages in the mapping,
1066 and the number of clean and dirty private pages in the mapping.
1067 "Referenced" indicates the amount of memory currently marked as
1068 referenced or accessed.
1069 "Anonymous" shows the amount of memory
1070 that does not belong to any file.
1071 "Swap" shows how much
1072 would-be-anonymous memory is also used, but out on swap.
1074 The "KernelPageSize" entry is the page size used by the kernel to back a VMA.
1075 This matches the size used by the MMU in the majority of cases.
1076 However, one counter-example occurs on PPC64 kernels
1077 whereby a kernel using 64K as a base page size may still use 4K
1078 pages for the MMU on older processors.
1079 To distinguish, this
1080 patch reports "MMUPageSize" as the page size used by the MMU.
1082 The "Locked" indicates whether the mapping is locked in memory
1085 "VmFlags" field represents the kernel flags associated with
1086 the particular virtual memory area in two letter encoded manner.
1087 The codes are the following:
1097 gd - stack segment growns down
1099 dw - disabled write to the mapped file
1100 lo - pages are locked in memory
1101 io - memory mapped I/O area
1102 sr - sequential read advise provided
1103 rr - random read advise provided
1104 dc - do not copy area on fork
1105 de - do not expand area on remapping
1106 ac - area is accountable
1107 nr - swap space is not reserved for the area
1108 ht - area uses huge tlb pages
1109 nl - non-linear mapping
1110 ar - architecture specific flag
1111 dd - do not include area into core dump
1112 sd - soft-dirty flag
1114 hg - huge page advise flag
1115 nh - no-huge page advise flag
1116 mg - mergable advise flag
1119 .IR /proc/[pid]/smaps
1120 file is present only if the
1121 .B CONFIG_PROC_PAGE_MONITOR
1122 kernel configuration option is enabled.
1124 .IR /proc/[pid]/stack " (since Linux 2.6.29)"
1125 .\" 2ec220e27f5040aec1e88901c1b6ea3d135787ad
1126 This file provides a symbolic trace of the function calls in this
1127 process's kernel stack.
1128 This file is provided only if the kernel was built with the
1129 .B CONFIG_STACKTRACE
1130 configuration option.
1133 Status information about the process.
1137 .IR /usr/src/linux/fs/proc/array.c "."
1139 The fields, in order, with their proper
1141 format specifiers, are:
1148 (2) The filename of the executable, in parentheses.
1149 This is visible whether or not the executable is swapped out.
1152 (3) One character from the string "RSDZTW" where R is running, S is
1153 sleeping in an interruptible wait, D is waiting in uninterruptible
1154 disk sleep, Z is zombie, T is traced or stopped (on a signal),
1158 (4) The PID of the parent.
1161 (5) The process group ID of the process.
1164 (6) The session ID of the process.
1167 (7) The controlling terminal of the process.
1168 (The minor device number is contained in the combination of bits
1169 31 to 20 and 7 to 0;
1170 the major device number is in bits 15 to 8.)
1173 .\" This field and following, up to and including wchan added 0.99.1
1174 (8) The ID of the foreground process group of the controlling
1175 terminal of the process.
1177 \fIflags\fP %u (%lu before Linux 2.6.22)
1178 (9) The kernel flags word of the process.
1180 see the PF_* defines in the Linux kernel source file
1181 .IR include/linux/sched.h .
1182 Details depend on the kernel version.
1185 (10) The number of minor faults the process has made which have not
1186 required loading a memory page from disk.
1189 (11) The number of minor faults that the process's
1190 waited-for children have made.
1193 (12) The number of major faults the process has made which have
1194 required loading a memory page from disk.
1197 (13) The number of major faults that the process's
1198 waited-for children have made.
1201 (14) Amount of time that this process has been scheduled in user mode,
1202 measured in clock ticks (divide by
1203 .IR sysconf(_SC_CLK_TCK) ).
1204 This includes guest time, \fIguest_time\fP
1205 (time spent running a virtual CPU, see below),
1206 so that applications that are not aware of the guest time field
1207 do not lose that time from their calculations.
1210 (15) Amount of time that this process has been scheduled in kernel mode,
1211 measured in clock ticks (divide by
1212 .IR sysconf(_SC_CLK_TCK) ).
1215 (16) Amount of time that this process's
1216 waited-for children have been scheduled in user mode,
1217 measured in clock ticks (divide by
1218 .IR sysconf(_SC_CLK_TCK) ).
1221 This includes guest time, \fIcguest_time\fP
1222 (time spent running a virtual CPU, see below).
1225 (17) Amount of time that this process's
1226 waited-for children have been scheduled in kernel mode,
1227 measured in clock ticks (divide by
1228 .IR sysconf(_SC_CLK_TCK) ).
1231 (18) (Explanation for Linux 2.6)
1232 For processes running a real-time scheduling policy
1235 .BR sched_setscheduler (2)),
1236 this is the negated scheduling priority, minus one;
1237 that is, a number in the range \-2 to \-100,
1238 corresponding to real-time priorities 1 to 99.
1239 For processes running under a non-real-time scheduling policy,
1240 this is the raw nice value
1241 .RB ( setpriority (2))
1242 as represented in the kernel.
1243 The kernel stores nice values as numbers
1244 in the range 0 (high) to 39 (low),
1245 corresponding to the user-visible nice range of \-20 to 19.
1247 Before Linux 2.6, this was a scaled value based on
1248 the scheduler weighting given to this process.
1249 .\" And back in kernel 1.2 days things were different again.
1252 (19) The nice value (see
1253 .BR setpriority (2)),
1254 a value in the range 19 (low priority) to \-20 (high priority).
1255 .\" Back in kernel 1.2 days things were different.
1257 .\" \fIcounter\fP %ld
1258 .\" The current maximum size in jiffies of the process's next timeslice,
1259 .\" or what is currently left of its current timeslice, if it is the
1260 .\" currently running process.
1262 .\" \fItimeout\fP %u
1263 .\" The time in jiffies of the process's next timeout.
1264 .\" timeout was removed sometime around 2.1/2.2
1266 \fInum_threads\fP %ld
1267 (20) Number of threads in this process (since Linux 2.6).
1268 Before kernel 2.6, this field was hard coded to 0 as a placeholder
1269 for an earlier removed field.
1271 \fIitrealvalue\fP %ld
1272 (21) The time in jiffies before the next
1274 is sent to the process due to an interval timer.
1275 Since kernel 2.6.17, this field is no longer maintained,
1276 and is hard coded as 0.
1278 \fIstarttime\fP %llu (was %lu before Linux 2.6)
1279 (22) The time the process started after system boot.
1280 In kernels before Linux 2.6, this value was expressed in jiffies.
1281 Since Linux 2.6, the value is expressed in clock ticks (divide by
1282 .IR sysconf(_SC_CLK_TCK) ).
1285 (23) Virtual memory size in bytes.
1288 (24) Resident Set Size: number of pages the process has in real memory.
1289 This is just the pages which
1290 count toward text, data, or stack space.
1291 This does not include pages
1292 which have not been demand-loaded in, or which are swapped out.
1295 (25) Current soft limit in bytes on the rss of the process;
1296 see the description of
1302 (26) The address above which program text can run.
1305 (27) The address below which program text can run.
1307 \fIstartstack\fP %lu
1308 (28) The address of the start (i.e., bottom) of the stack.
1311 (29) The current value of ESP (stack pointer), as found in the
1312 kernel stack page for the process.
1315 (30) The current EIP (instruction pointer).
1318 (31) The bitmap of pending signals, displayed as a decimal number.
1319 Obsolete, because it does not provide information on real-time signals; use
1320 .I /proc/[pid]/status
1324 (32) The bitmap of blocked signals, displayed as a decimal number.
1325 Obsolete, because it does not provide information on real-time signals; use
1326 .I /proc/[pid]/status
1330 (33) The bitmap of ignored signals, displayed as a decimal number.
1331 Obsolete, because it does not provide information on real-time signals; use
1332 .I /proc/[pid]/status
1336 (34) The bitmap of caught signals, displayed as a decimal number.
1337 Obsolete, because it does not provide information on real-time signals; use
1338 .I /proc/[pid]/status
1342 (35) This is the "channel" in which the process is waiting.
1343 It is the address of a location in the kernel where the process is sleeping.
1344 The corresponding symbolic name can be found in
1345 .IR /proc/[pid]/wchan .
1349 .\" nswap was added in 2.0
1350 Number of pages swapped (not maintained).
1354 .\" cnswap was added in 2.0
1355 Cumulative \fInswap\fP for child processes (not maintained).
1357 \fIexit_signal\fP %d (since Linux 2.1.22)
1358 (38) Signal to be sent to parent when we die.
1360 \fIprocessor\fP %d (since Linux 2.2.8)
1361 (39) CPU number last executed on.
1363 \fIrt_priority\fP %u (since Linux 2.5.19; was %lu before Linux 2.6.22)
1364 (40) Real-time scheduling priority, a number in the range 1 to 99 for
1365 processes scheduled under a real-time policy,
1366 or 0, for non-real-time processes (see
1367 .BR sched_setscheduler (2)).
1369 \fIpolicy\fP %u (since Linux 2.5.19; was %lu before Linux 2.6.22)
1370 (41) Scheduling policy (see
1371 .BR sched_setscheduler (2)).
1372 Decode using the SCHED_* constants in
1375 \fIdelayacct_blkio_ticks\fP %llu (since Linux 2.6.18)
1376 (42) Aggregated block I/O delays, measured in clock ticks (centiseconds).
1378 \fIguest_time\fP %lu (since Linux 2.6.24)
1379 (43) Guest time of the process (time spent running a virtual CPU
1380 for a guest operating system), measured in clock ticks (divide by
1381 .IR sysconf(_SC_CLK_TCK) ).
1383 \fIcguest_time\fP %ld (since Linux 2.6.24)
1384 (44) Guest time of the process's children, measured in clock ticks (divide by
1385 .IR sysconf(_SC_CLK_TCK) ).
1388 .I /proc/[pid]/statm
1389 Provides information about memory usage, measured in pages.
1394 size (1) total program size
1395 (same as VmSize in \fI/proc/[pid]/status\fP)
1396 resident (2) resident set size
1397 (same as VmRSS in \fI/proc/[pid]/status\fP)
1398 share (3) shared pages (i.e., backed by a file)
1399 text (4) text (code)
1400 .\" (not including libs; broken, includes data segment)
1401 lib (5) library (unused in Linux 2.6)
1402 data (6) data + stack
1403 .\" (including libs; broken, includes library text)
1404 dt (7) dirty pages (unused in Linux 2.6)
1408 .I /proc/[pid]/status
1409 Provides much of the information in
1412 .I /proc/[pid]/statm
1413 in a format that's easier for humans to parse.
1418 .RB "$" " cat /proc/$$/status"
1425 Uid: 1000 1000 1000 1000
1426 Gid: 100 100 100 100
1441 SigPnd: 0000000000000000
1442 ShdPnd: 0000000000000000
1443 SigBlk: 0000000000010000
1444 SigIgn: 0000000000384004
1445 SigCgt: 000000004b813efb
1446 CapInh: 0000000000000000
1447 CapPrm: 0000000000000000
1448 CapEff: 0000000000000000
1449 CapBnd: ffffffffffffffff
1450 Cpus_allowed: 00000001
1451 Cpus_allowed_list: 0
1453 Mems_allowed_list: 0
1454 voluntary_ctxt_switches: 150
1455 nonvoluntary_ctxt_switches: 545
1459 The fields are as follows:
1463 Command run by this process.
1466 Current state of the process.
1478 Thread group ID (i.e., Process ID).
1485 PID of parent process.
1488 PID of process tracing this process (0 if not being traced).
1491 Real, effective, saved set, and filesystem UIDs (GIDs).
1494 Number of file descriptor slots currently allocated.
1497 Supplementary group list.
1500 Peak virtual memory size.
1503 Virtual memory size.
1506 Locked memory size (see
1510 Peak resident set size ("high water mark").
1515 .IR VmData ", " VmStk ", " VmExe :
1516 Size of data, stack, and text segments.
1519 Shared library code size.
1522 Page table entries size (since Linux 2.6.10).
1525 Number of threads in process containing this thread.
1528 This field contains two slash-separated numbers that relate to
1529 queued signals for the real user ID of this process.
1530 The first of these is the number of currently queued
1531 signals for this real user ID, and the second is the
1532 resource limit on the number of queued signals for this process
1533 (see the description of
1534 .BR RLIMIT_SIGPENDING
1538 .IR SigPnd ", " ShdPnd :
1539 Number of signals pending for thread and for process as a whole (see
1544 .IR SigBlk ", " SigIgn ", " SigCgt :
1545 Masks indicating signals being blocked, ignored, and caught (see
1548 .IR CapInh ", " CapPrm ", " CapEff :
1549 Masks of capabilities enabled in inheritable, permitted, and effective sets
1551 .BR capabilities (7)).
1554 Capability Bounding set
1555 (since kernel 2.6.26, see
1556 .BR capabilities (7)).
1559 Mask of CPUs on which this process may run
1560 (since Linux 2.6.24, see
1563 .IR Cpus_allowed_list :
1564 Same as previous, but in "list format"
1565 (since Linux 2.6.26, see
1569 Mask of memory nodes allowed to this process
1570 (since Linux 2.6.24, see
1573 .IR Mems_allowed_list :
1574 Same as previous, but in "list format"
1575 (since Linux 2.6.26, see
1578 .IR voluntary_context_switches ", " nonvoluntary_context_switches :
1579 Number of voluntary and involuntary context switches (since Linux 2.6.23).
1582 .IR /proc/[pid]/task " (since Linux 2.6.0-test6)"
1583 This is a directory that contains one subdirectory
1584 for each thread in the process.
1585 The name of each subdirectory is the numerical thread ID
1589 Within each of these subdirectories, there is a set of
1590 files with the same names and contents as under the
1593 For attributes that are shared by all threads, the contents for
1594 each of the files under the
1596 subdirectories will be the same as in the corresponding
1600 (e.g., in a multithreaded process, all of the
1602 files will have the same value as the
1604 file in the parent directory, since all of the threads in a process
1605 share a working directory).
1606 For attributes that are distinct for each thread,
1607 the corresponding files under
1609 may have different values (e.g., various fields in each of the
1610 .I task/[tid]/status
1611 files may be different for each thread).
1613 .\" The following was still true as at kernel 2.6.13
1614 In a multithreaded process, the contents of the
1616 directory are not available if the main thread has already terminated
1617 (typically by calling
1618 .BR pthread_exit (3)).
1620 .IR /proc/[pid]/wchan " (since Linux 2.6.0)"
1621 The symbolic name corresponding to the location
1622 in the kernel where the process is sleeping.
1625 Advanced power management version and battery information when
1627 is defined at kernel compilation time.
1630 Contains subdirectories for installed busses.
1633 Subdirectory for PCMCIA devices when
1635 is set at kernel compilation time.
1637 .IR /proc/[pid]/timers " (since Linux 3.10)"
1638 .\" commit 5ed67f05f66c41e39880a6d61358438a25f9fee5
1639 .\" commit 48f6a7a511ef8823fdff39afee0320092d43a8a0
1640 A list of the POSIX timers for this process.
1641 Each timer is listed with a line that started with the string "ID:".
1647 signal: 60/00007fff86e452a8
1648 notify: signal/pid.2634
1651 signal: 60/00007fff86e452a8
1652 notify: signal/pid.2634
1657 The lines shown for each timer have the following meanings:
1661 The ID for this timer.
1662 This is not the same as the timer ID returned by
1663 .BR timer_create (2);
1664 rather, it is the same kernel-internal ID that is available via the
1672 This is the signal number that this timer uses to deliver notifications
1673 followed by a slash, and then the
1674 .I sigev_value.sival_ptr
1675 value supplied to the signal handler.
1676 Valid only for timers that notify via a signal.
1679 The part before the slash specifies the mechanism
1680 that this timer uses to deliver notifications,
1681 and is one of "thread", "signal", or "none".
1682 Immediately following the slash is either the string "tid" for timers
1685 notification, or "pid" for timers that notify by other mechanisms.
1686 Following the "." is the PID of the process that will be delivered
1687 a signal if the timer delivers notifications via a signal.
1690 This field identifies the clock that the timer uses for measuring time.
1691 For most clocks, this is a number that matches one of the user-space
1693 constants exposed via
1695 .B CLOCK_PROCESS_CPUTIME_ID
1696 timers display with a value of -6
1698 .B CLOCK_THREAD_CPUTIME_ID
1699 timers display with a value of -2
1703 .I /proc/bus/pccard/drivers
1706 Contains various bus subdirectories and pseudo-files containing
1707 information about PCI busses, installed devices, and device
1709 Some of these files are not ASCII.
1711 .I /proc/bus/pci/devices
1712 Information about PCI devices.
1713 They may be accessed through
1719 Arguments passed to the Linux kernel at boot time.
1720 Often done via a boot manager such as
1725 .IR /proc/config.gz " (since Linux 2.6)"
1726 This file exposes the configuration options that were used
1727 to build the currently running kernel,
1728 in the same format as they would be shown in the
1730 file that resulted when configuring the kernel (using
1731 .IR "make xconfig" ,
1734 The file contents are compressed; view or search them using
1738 As long as no changes have been made to the following file,
1741 are the same as those provided by :
1745 cat /lib/modules/$(uname \-r)/build/.config
1750 is provided only if the kernel is configured with
1751 .BR CONFIG_IKCONFIG_PROC .
1754 This is a collection of CPU and system architecture dependent items,
1755 for each supported architecture a different list.
1756 Two common entries are \fIprocessor\fP which gives CPU number and
1757 \fIbogomips\fP; a system constant that is calculated
1758 during kernel initialization.
1759 SMP machines have information for
1763 command gathers its information from this file.
1766 Text listing of major numbers and device groups.
1767 This can be used by MAKEDEV scripts for consistency with the kernel.
1769 .IR /proc/diskstats " (since Linux 2.5.69)"
1770 This file contains disk I/O statistics for each disk device.
1771 See the Linux kernel source file
1772 .I Documentation/iostats.txt
1773 for further information.
1776 This is a list of the registered \fIISA\fP DMA (direct memory access)
1782 .I /proc/execdomains
1783 List of the execution domains (ABI personalities).
1786 Frame buffer information when
1788 is defined during kernel compilation.
1790 .I /proc/filesystems
1791 A text listing of the filesystems which are supported by the kernel,
1792 namely filesystems which were compiled into the kernel or whose kernel
1793 modules are currently loaded.
1795 .BR filesystems (5).)
1796 If a filesystem is marked with "nodev",
1797 this means that it does not require a block device to be mounted
1798 (e.g., virtual filesystem, network filesystem).
1800 Incidentally, this file may be used by
1802 when no filesystem is specified and it didn't manage to determine the
1804 Then filesystems contained in this file are tried
1805 (excepted those that are marked with "nodev").
1812 exists on systems with the IDE bus.
1813 There are directories for each IDE channel and attached device.
1818 cache buffer size in KB
1819 capacity number of sectors
1820 driver driver version
1821 geometry physical and logical geometry
1822 identify in hexadecimal
1824 model manufacturer's model number
1825 settings drive settings
1826 smart_thresholds in hexadecimal
1827 smart_values in hexadecimal
1833 utility provides access to this information in a friendly format.
1836 This is used to record the number of interrupts per CPU per IO device.
1838 for the i386 and x86_64 architectures, at least, this also includes
1839 interrupts internal to the system (that is, not associated with a device
1840 as such), such as NMI (nonmaskable interrupt), LOC (local timer interrupt),
1841 and for SMP systems, TLB (TLB flush interrupt), RES (rescheduling
1842 interrupt), CAL (remote function call interrupt), and possibly others.
1843 Very easy to read formatting, done in ASCII.
1846 I/O memory map in Linux 2.4.
1849 This is a list of currently registered Input-Output port regions that
1852 .IR /proc/kallsyms " (since Linux 2.5.71)"
1853 This holds the kernel exported symbol definitions used by the
1855 tools to dynamically link and bind loadable modules.
1856 In Linux 2.5.47 and earlier, a similar file with slightly different syntax
1861 This file represents the physical memory of the system and is stored
1862 in the ELF core file format.
1863 With this pseudo-file, and an unstripped
1865 .RI ( /usr/src/linux/vmlinux )
1866 binary, GDB can be used to
1867 examine the current state of any kernel data structures.
1869 The total length of the file is the size of physical memory (RAM) plus
1873 This file can be used instead of the
1875 system call to read kernel messages.
1876 A process must have superuser
1877 privileges to read this file, and only one process should read this
1879 This file should not be read if a syslog process is running
1882 system call facility to log kernel messages.
1884 Information in this file is retrieved with the
1888 .IR /proc/ksyms " (Linux 1.1.23-2.5.47)"
1890 .IR /proc/kallsyms .
1893 The first three fields in this file are load average figures
1894 giving the number of jobs in the run queue (state R)
1895 or waiting for disk I/O (state D) averaged over 1, 5, and 15 minutes.
1896 They are the same as the load average numbers given by
1899 The fourth field consists of two numbers separated by a slash (/).
1900 The first of these is the number of currently runnable kernel
1901 scheduling entities (processes, threads).
1902 The value after the slash is the number of kernel scheduling entities
1903 that currently exist on the system.
1904 The fifth field is the PID of the process that was most
1905 recently created on the system.
1908 This file shows current file locks
1909 .RB ( flock "(2) and " fcntl (2))
1913 .IR /proc/malloc " (only up to and including Linux 2.2)"
1914 .\" It looks like this only ever did something back in 1.0 days
1915 This file is present only if
1916 .B CONFIG_DEBUG_MALLOC
1917 was defined during compilation.
1920 This file reports statistics about memory usage on the system.
1923 to report the amount of free and used memory (both physical and swap)
1924 on the system as well as the shared memory and buffers used by the
1926 Each line of the file consists of a parameter name, followed by a colon,
1927 the value of the parameter, and an option unit of measurement (e.g., "kB").
1928 The list below describes the parameter names and
1929 the format specifier required to read the field value.
1930 Except as noted below,
1931 all of the fields have been present since at least Linux 2.6.0.
1932 Some fileds are displayed only if the kernel was configured
1933 with various options; those dependencies are noted in the list.
1937 Total usable RAM (i.e., physical RAM minus a few reserved
1938 bits and the kernel binary code).
1942 .IR LowFree + HighFree .
1945 Relatively temporary storage for raw disk blocks that
1946 shouldn't get tremendously large (20MB or so).
1949 In-memory cache for files read from the disk (the page cache).
1953 .IR SwapCached " %lu"
1954 Memory that once was swapped out, is swapped back in but
1955 still also is in the swap file.
1956 (If memory pressure is high, these pages
1957 don't need to be swapped out again because they are already
1962 Memory that has been used more recently and usually not
1963 reclaimed unless absolutely necessary.
1966 Memory which has been less recently used.
1967 It is more eligible to be reclaimed for other purposes.
1969 .IR Active(anon) " %lu (since Linux 2.6.28)"
1972 .IR Inactive(anon) " %lu (since Linux 2.6.28)"
1975 .IR Active(file) " %lu (since Linux 2.6.28)"
1978 .IR Inactive(file) " %lu (since Linux 2.6.28)"
1981 .IR Unevictable " %lu (since Linux 2.6.28)"
1982 (From Linux 2.6.28 to 2.6.30,
1983 \fBCONFIG_UNEVICTABLE_LRU\fP was required.)
1986 .IR Mlocked " %lu (since Linux 2.6.28)"
1987 (From Linux 2.6.28 to 2.6.30,
1988 \fBCONFIG_UNEVICTABLE_LRU\fP was required.)
1991 .IR HighTotal " %lu"
1992 (Starting with Linux 2.6.19, \fBCONFIG_HIGHMEM\fP is required.)
1993 Total amount of highmem.
1994 Highmem is all memory above ~860MB of physical memory.
1995 Highmem areas are for use by user-space programs,
1996 or for the page cache.
1997 The kernel must use tricks to access
1998 this memory, making it slower to access than lowmem.
2001 (Starting with Linux 2.6.19, \fBCONFIG_HIGHMEM\fP is required.)
2002 Amount of free highmem.
2005 (Starting with Linux 2.6.19, \fBCONFIG_HIGHMEM\fP is required.)
2006 Total amount of lowmem.
2007 Lowmem is memory which can be used for everything that
2008 highmem can be used for, but it is also available for the
2009 kernel's use for its own data structures.
2010 Among many other things,
2011 it is where everything from
2014 Bad things happen when you're out of lowmem.
2017 (Starting with Linux 2.6.19, \fBCONFIG_HIGHMEM\fP is required.)
2018 Amount of free lowmem.
2020 .IR MmapCopy " %lu (since Linux 2.6.29)"
2025 .IR SwapTotal " %lu"
2026 Total amount of swap space available.
2029 Amount of swap space that is currently unused.
2032 Memory which is waiting to get written back to the disk.
2034 .IR Writeback " %lu"
2035 Memory which is actively being written back to the disk.
2037 .IR AnonPages " %lu (since Linux 2.6.18)"
2038 Non-file backed pages mapped into user-space page tables.
2041 Files which have been mmaped, such as libraries.
2043 .IR Shmem " %lu (since Linux 2.6.32)"
2047 In-kernel data structures cache.
2049 .IR SReclaimable " %lu (since Linux 2.6.19)"
2052 that might be reclaimed, such as caches.
2054 .IR SUnreclaim " %lu (since Linux 2.6.19)"
2057 that cannot be reclaimed on memory pressure.
2059 .IR KernelStack " %lu (since Linux 2.6.32)"
2060 Amount of memory allocated to kernel stacks.
2062 .IR PageTables " %lu (since Linux 2.6.18)"
2063 Amount of memory dedicated to the lowest level of page tables.
2065 .IR Quicklists " %lu (since Linux 2.6.27)"
2066 (\fBCONFIG_QUICKLIST\fP is required.)
2069 .IR NFS_Unstable " %lu (since Linux 2.6.18)"
2070 NFS pages sent to the server, but not yet committed to stable storage.
2072 .IR Bounce " %lu (since Linux 2.6.18)"
2073 Memory used for block device "bounce buffers".
2075 .IR WritebackTmp " %lu (since Linux 2.6.26)"
2076 Memory used by FUSE for temporary writeback buffers.
2078 .IR CommitLimit " %lu (since Linux 2.6.10)"
2079 Based on the overcommit ratio ('vm.overcommit_ratio'),
2080 this is the total amount of memory currently available to
2081 be allocated on the system.
2082 This limit is adhered to
2083 only if strict overcommit accounting is enabled (mode 2 in
2084 .IR /proc/sys/vm/overcommit_ratio ).
2087 is calculated using the following formula:
2090 ([total RAM pages] - [total huge TLB pages]) *
2091 overcommit_ratio / 100 + [total swap pages]
2093 For example, on a system with 1GB of physical RAM and 7GB
2096 of 30, this formula yields a
2099 For more details, see the memory overcommit documentation
2100 in the kernel source file
2101 .IR Documentation/vm/overcommit-accounting .
2103 .IR Committed_AS " %lu"
2104 The amount of memory presently allocated on the system.
2105 The committed memory is a sum of all of the memory which
2106 has been allocated by processes, even if it has not been
2107 "used" by them as of yet.
2108 A process which allocates 1GB of memory (using
2110 or similar), but touches only 300MB of that memory will show up
2111 as using only 300MB of memory even if it has the address space
2112 allocated for the entire 1GB.
2113 This 1GB is memory which has been "committed" to by the VM
2114 and can be used at any time by the allocating application.
2115 With strict overcommit enabled on the system (mode 2
2116 .IR /proc/sys/vm/overcommit_memory ),
2117 allocations which would exceed the
2119 (detailed above) will not be permitted.
2120 This is useful if one needs to guarantee that processes will not
2121 fail due to lack of memory once that memory has been successfully allocated.
2123 .IR VmallocTotal " %lu"
2124 Total size of vmalloc memory area.
2126 .IR VmallocUsed " %lu"
2127 Amount of vmalloc area which is used.
2129 .IR VmallocChunk " %lu"
2130 Largest contiguous block of vmalloc area which is free.
2132 .IR HardwareCorrupted " %lu (since Linux 2.6.32)"
2133 (\fBCONFIG_MEMORY_FAILURE\fP is required.)
2136 .IR AnonHugePages " %lu (since Linux 2.6.38)"
2137 (\fBCONFIG_TRANSPARENT_HUGEPAGE\fP is required.)
2138 Non-file backed huge pages mapped into user-space page tables.
2140 .IR HugePages_Total " %lu"
2141 (\fBCONFIG_HUGETLB_PAGE\fP is required.)
2142 The size of the pool of huge pages.
2144 .IR HugePages_Free " %lu"
2145 (\fBCONFIG_HUGETLB_PAGE\fP is required.)
2146 The number of huge pages in the pool that are not yet allocated.
2148 .IR HugePages_Rsvd " %lu (since Linux 2.6.17)"
2149 (\fBCONFIG_HUGETLB_PAGE\fP is required.)
2150 This is the number of huge pages for
2151 which a commitment to allocate from the pool has been made,
2152 but no allocation has yet been made.
2153 These reserved huge pages
2154 guarantee that an application will be able to allocate a
2155 huge page from the pool of huge pages at fault time.
2157 .IR HugePages_Surp " %lu (since Linux 2.6.24)"
2158 (\fBCONFIG_HUGETLB_PAGE\fP is required.)
2159 This is the number of huge pages in
2160 the pool above the value in
2161 .IR /proc/sys/vm/nr_hugepages .
2162 The maximum number of surplus huge pages is controlled by
2163 .IR /proc/sys/vm/nr_overcommit_hugepages .
2165 .IR Hugepagesize " %lu"
2166 (\fBCONFIG_HUGETLB_PAGE\fP is required.)
2167 The size of huge pages.
2171 A text list of the modules that have been loaded by the system.
2176 Before kernel 2.4.19, this file was a list
2177 of all the filesystems currently mounted on the system.
2178 With the introduction of per-process mount namespaces in
2179 Linux 2.4.19, this file became a link to
2180 .IR /proc/self/mounts ,
2181 which lists the mount points of the process's own mount namespace.
2182 The format of this file is documented in
2186 Memory Type Range Registers.
2187 See the Linux kernel source file
2188 .I Documentation/mtrr.txt
2192 various net pseudo-files, all of which give the status of some part of
2193 the networking layer.
2194 These files contain ASCII structures and are,
2195 therefore, readable with
2197 However, the standard
2199 suite provides much cleaner access to these files.
2202 This holds an ASCII readable dump of the kernel ARP table used for
2203 address resolutions.
2204 It will show both dynamically learned and preprogrammed ARP entries.
2210 IP address HW type Flags HW address Mask Device
2211 192.168.0.50 0x1 0x2 00:50:BF:25:68:F3 * eth0
2212 192.168.0.250 0x1 0xc 00:00:00:00:00:00 * eth0
2217 Here "IP address" is the IPv4 address of the machine and the "HW type"
2218 is the hardware type of the address from RFC\ 826.
2219 The flags are the internal
2220 flags of the ARP structure (as defined in
2221 .IR /usr/include/linux/if_arp.h )
2223 the "HW address" is the data link layer mapping for that IP address if
2227 The dev pseudo-file contains network device status information.
2229 the number of received and sent packets, the number of errors and
2231 and other basic statistics.
2232 These are used by the
2234 program to report device status.
2240 Inter-| Receive | Transmit
2241 face |bytes packets errs drop fifo frame compressed multicast|bytes packets errs drop fifo colls carrier compressed
2242 lo: 2776770 11307 0 0 0 0 0 0 2776770 11307 0 0 0 0 0 0
2243 eth0: 1215645 2751 0 0 0 0 0 0 1782404 4324 0 0 0 427 0 0
2244 ppp0: 1622270 5552 1 0 0 0 0 0 354130 5669 0 0 0 0 0 0
2245 tap0: 7714 81 0 0 0 0 0 0 7714 81 0 0 0 0 0 0
2250 .\" .I /proc/net/ipx
2253 .\" .I /proc/net/ipx_route
2256 .I /proc/net/dev_mcast
2258 .IR /usr/src/linux/net/core/dev_mcast.c :
2261 indx interface_name dmi_u dmi_g dmi_address
2262 2 eth0 1 0 01005e000001
2263 3 eth1 1 0 01005e000001
2264 4 eth2 1 0 01005e000001
2269 Internet Group Management Protocol.
2271 .IR /usr/src/linux/net/core/igmp.c .
2274 This file uses the same format as the
2276 file and contains the current reverse mapping database used to provide
2278 reverse address lookup services.
2279 If RARP is not configured into the
2281 this file will not be present.
2284 Holds a dump of the RAW socket table.
2285 Much of the information is not of
2287 apart from debugging.
2288 The "sl" value is the kernel hash slot for the
2290 the "local_address" is the local address and protocol number pair.
2292 the internal status of the socket.
2293 The "tx_queue" and "rx_queue" are the
2294 outgoing and incoming data queue in terms of kernel memory usage.
2295 The "tr", "tm\->when", and "rexmits" fields are not used by RAW.
2297 field holds the effective UID of the creator of the socket.
2299 .\" .I /proc/net/route
2300 .\" No information, but looks similar to
2304 This file holds the ASCII data needed for the IP, ICMP, TCP, and UDP
2306 information bases for an SNMP agent.
2309 Holds a dump of the TCP socket table.
2310 Much of the information is not
2311 of use apart from debugging.
2312 The "sl" value is the kernel hash slot
2313 for the socket, the "local_address" is the local address and port number pair.
2314 The "rem_address" is the remote address and port number pair
2316 \&"St" is the internal status of the socket.
2317 The "tx_queue" and "rx_queue" are the
2318 outgoing and incoming data queue in terms of kernel memory usage.
2319 The "tr", "tm\->when", and "rexmits" fields hold internal information of
2320 the kernel socket state and are only useful for debugging.
2322 field holds the effective UID of the creator of the socket.
2325 Holds a dump of the UDP socket table.
2326 Much of the information is not of
2327 use apart from debugging.
2328 The "sl" value is the kernel hash slot for the
2329 socket, the "local_address" is the local address and port number pair.
2330 The "rem_address" is the remote address and port number pair
2331 (if connected). "St" is the internal status of the socket.
2332 The "tx_queue" and "rx_queue" are the outgoing and incoming data queue
2333 in terms of kernel memory usage.
2334 The "tr", "tm\->when", and "rexmits" fields
2335 are not used by UDP.
2337 field holds the effective UID of the creator of the socket.
2343 sl local_address rem_address st tx_queue rx_queue tr rexmits tm\->when uid
2344 1: 01642C89:0201 0C642C89:03FF 01 00000000:00000001 01:000071BA 00000000 0
2345 1: 00000000:0801 00000000:0000 0A 00000000:00000000 00:00000000 6F000100 0
2346 1: 00000000:0201 00000000:0000 0A 00000000:00000000 00:00000000 00000000 0
2352 Lists the UNIX domain sockets present within the system and their
2358 Num RefCount Protocol Flags Type St Path
2359 0: 00000002 00000000 00000000 0001 03
2360 1: 00000001 00000000 00010000 0001 01 /dev/printer
2365 Here "Num" is the kernel table slot number, "RefCount" is the number
2366 of users of the socket, "Protocol" is currently always 0, "Flags"
2367 represent the internal kernel flags holding the status of the
2369 Currently, type is always "1" (UNIX domain datagram sockets are
2370 not yet supported in the kernel).
2371 \&"St" is the internal state of the
2372 socket and Path is the bound path (if any) of the socket.
2375 Contains the major and minor numbers of each partition as well as the number
2376 of 1024-byte blocks and the partition name.
2379 This is a listing of all PCI devices found during kernel initialization
2380 and their configuration.
2382 This file has been deprecated in favor of a new
2385 .RI ( /proc/bus/pci ).
2386 It became optional in Linux 2.2 (available with
2387 .B CONFIG_PCI_OLD_PROC
2388 set at kernel compilation).
2389 It became once more nonoptionally enabled in Linux 2.4.
2390 Next, it was deprecated in Linux 2.6 (still available with
2391 .B CONFIG_PCI_LEGACY_PROC
2392 set), and finally removed altogether since Linux 2.6.17.
2393 .\" FIXME /proc/sched_debug
2395 .\" .IR /proc/sched_debug " (since Linux 2.6.23)"
2396 .\" See also /proc/[pid]/sched
2398 .IR /proc/profile " (since Linux 2.4)"
2399 This file is present only if the kernel was booted with the
2401 command-line option.
2402 It exposes kernel profiling information in a binary format for use by
2403 .BR readprofile (1).
2404 Writing (e.g., an empty string) to this file resets the profiling counters;
2405 on some architectures,
2406 writing a binary integer "profiling multiplier" of size
2408 sets the profiling interrupt frequency.
2411 A directory with the
2413 mid-level pseudo-file and various SCSI low-level
2415 which contain a file for each SCSI host in this system, all of
2416 which give the status of some part of the SCSI IO subsystem.
2417 These files contain ASCII structures and are, therefore, readable with
2420 You can also write to some of the files to reconfigure the subsystem or
2421 switch certain features on or off.
2424 This is a listing of all SCSI devices known to the kernel.
2425 The listing is similar to the one seen during bootup.
2426 scsi currently supports only the \fIadd-single-device\fP command which
2427 allows root to add a hotplugged device to the list of known devices.
2433 echo \(aqscsi add-single-device 1 0 5 0\(aq > /proc/scsi/scsi
2438 host scsi1 to scan on SCSI channel 0 for a device on ID 5 LUN 0.
2440 is already a device known on this address or the address is invalid, an
2441 error will be returned.
2443 .I /proc/scsi/[drivername]
2444 \fI[drivername]\fP can currently be NCR53c7xx, aha152x, aha1542, aha1740,
2445 aic7xxx, buslogic, eata_dma, eata_pio, fdomain, in2000, pas16, qlogic,
2446 scsi_debug, seagate, t128, u15-24f, ultrastore, or wd7000.
2447 These directories show up for all drivers that registered at least one
2449 Every directory contains one file per registered host.
2450 Every host-file is named after the number the host was assigned during
2453 Reading these files will usually show driver and host configuration,
2454 statistics, and so on.
2456 Writing to these files allows different things on different hosts.
2457 For example, with the \fIlatency\fP and \fInolatency\fP commands,
2458 root can switch on and off command latency measurement code in the
2460 With the \fIlockup\fP and \fIunlock\fP commands,
2461 root can control bus lockups simulated by the scsi_debug driver.
2464 This directory refers to the process accessing the
2467 and is identical to the
2469 directory named by the process ID of the same process.
2472 Information about kernel caches.
2473 Since Linux 2.6.16 this file is present only if the
2475 kernel configuration option is enabled.
2497 kernel/system statistics.
2498 Varies with architecture.
2503 \fIcpu 3357 0 4313 1362393\fP
2504 The amount of time, measured in units of
2505 USER_HZ (1/100ths of a second on most architectures, use
2506 .IR sysconf(_SC_CLK_TCK)
2507 to obtain the right value),
2508 .\" 1024 on Alpha and ia64
2509 that the system spent in various states:
2513 (1) Time spent in user mode.
2516 (2) Time spent in user mode with low priority (nice).
2519 (3) Time spent in system mode.
2522 (4) Time spent in the idle task.
2523 .\" FIXME Actually, the following info about the /proc/stat 'cpu' field
2524 .\" does not seem to be quite right (at least in 2.6.12 or 3.6):
2525 .\" the idle time in /proc/uptime does not quite match this value
2526 This value should be USER_HZ times the
2531 .IR iowait " (since Linux 2.5.41)"
2532 (5) Time waiting for I/O to complete.
2534 .IR irq " (since Linux 2.6.0-test4)"
2535 (6) Time servicing interrupts.
2537 .IR softirq " (since Linux 2.6.0-test4)"
2538 (7) Time servicing softirqs.
2540 .IR steal " (since Linux 2.6.11)"
2541 (8) Stolen time, which is the time spent in other operating systems when
2542 running in a virtualized environment
2544 .IR guest " (since Linux 2.6.24)"
2545 (9) Time spent running a virtual CPU for guest
2546 operating systems under the control of the Linux kernel.
2547 .\" See Changelog entry for 5e84cfde51cf303d368fcb48f22059f37b3872de
2549 .IR guest_nice " (since Linux 2.6.33)"
2550 .\" commit ce0e7b28fb75cb003cfc8d0238613aaf1c55e797
2551 (10) Time spent running a niced guest (virtual CPU for guest
2552 operating systems under the control of the Linux kernel).
2555 \fIpage 5741 1808\fP
2556 The number of pages the system paged in and the number that were paged
2560 The number of swap pages that have been brought in and out.
2562 .\" FIXME The following is not the full picture for the 'intr' of
2563 .\" /proc/stat on 2.6:
2565 This line shows counts of interrupts serviced since boot time,
2566 for each of the possible system interrupts.
2567 The first column is the total of all interrupts serviced;
2568 each subsequent column is the total for a particular interrupt.
2570 \fIdisk_io: (2,0):(31,30,5764,1,2) (3,0):\fP...
2571 (major,disk_idx):(noinfo, read_io_ops, blks_read, write_io_ops, blks_written)
2576 The number of context switches that the system underwent.
2578 \fIbtime 769041601\fP
2579 boot time, in seconds since the Epoch, 1970-01-01 00:00:00 +0000 (UTC).
2581 \fIprocesses 86031\fP
2582 Number of forks since boot.
2584 \fIprocs_running 6\fP
2585 Number of processes in runnable state.
2586 (Linux 2.5.45 onward.)
2588 \fIprocs_blocked 2\fP
2589 Number of processes blocked waiting for I/O to complete.
2590 (Linux 2.5.45 onward.)
2599 This directory (present since 1.3.57) contains a number of files
2600 and subdirectories corresponding to kernel variables.
2601 These variables can be read and sometimes modified using
2602 the \fI/proc\fP filesystem, and the (deprecated)
2606 .IR /proc/sys/abi " (since Linux 2.4.10)"
2607 This directory may contain files with application binary information.
2608 .\" On some systems, it is not present.
2609 See the Linux kernel source file
2610 .I Documentation/sysctl/abi.txt
2611 for more information.
2614 This directory may be empty.
2617 This directory contains device-specific information (e.g.,
2618 .IR dev/cdrom/info ).
2620 some systems, it may be empty.
2623 This directory contains the files and subdirectories for kernel variables
2624 related to filesystems.
2626 .I /proc/sys/fs/binfmt_misc
2627 Documentation for files in this directory can be found
2628 in the Linux kernel sources in
2629 .IR Documentation/binfmt_misc.txt .
2631 .IR /proc/sys/fs/dentry-state " (since Linux 2.2)"
2632 This file contains information about the status of the
2633 directory cache (dcache).
2634 The file contains six numbers,
2635 .IR nr_dentry ", " nr_unused ", " age_limit " (age in seconds), "
2637 (pages requested by system) and two dummy values.
2641 is the number of allocated dentries (dcache entries).
2642 This field is unused in Linux 2.2.
2645 is the number of unused dentries.
2648 .\" looks like this is unused in kernels 2.2 to 2.6
2649 is the age in seconds after which dcache entries
2650 can be reclaimed when memory is short.
2653 .\" looks like this is unused in kernels 2.2 to 2.6
2654 is nonzero when the kernel has called shrink_dcache_pages() and the
2655 dcache isn't pruned yet.
2658 .I /proc/sys/fs/dir-notify-enable
2659 This file can be used to disable or enable the
2661 interface described in
2663 on a system-wide basis.
2664 A value of 0 in this file disables the interface,
2665 and a value of 1 enables it.
2667 .I /proc/sys/fs/dquot-max
2668 This file shows the maximum number of cached disk quota entries.
2669 On some (2.4) systems, it is not present.
2670 If the number of free cached disk quota entries is very low and
2671 you have some awesome number of simultaneous system users,
2672 you might want to raise the limit.
2674 .I /proc/sys/fs/dquot-nr
2675 This file shows the number of allocated disk quota
2676 entries and the number of free disk quota entries.
2678 .IR /proc/sys/fs/epoll " (since Linux 2.6.28)"
2679 This directory contains the file
2680 .IR max_user_watches ,
2681 which can be used to limit the amount of kernel memory consumed by the
2684 For further details, see
2687 .I /proc/sys/fs/file-max
2689 a system-wide limit on the number of open files for all processes.
2692 which can be used by a process to set the per-process limit,
2694 on the number of files it may open.)
2696 of error messages in the kernel log about running out of file handles
2697 (look for "VFS: file-max limit <number> reached"),
2698 try increasing this value:
2704 echo 100000 > /proc/sys/fs/file-max
2710 imposes an upper limit on the value that may be placed in
2713 Privileged processes
2714 .RB ( CAP_SYS_ADMIN )
2719 .I /proc/sys/fs/file-nr
2720 This (read-only) file contains three numbers:
2721 the number of allocated file handles
2722 (i.e., the number of files presently opened);
2723 the number of free file handles;
2724 and the maximum number of file handles (i.e., the same value as
2725 .IR /proc/sys/fs/file-max ).
2726 If the number of allocated file handles is close to the
2727 maximum, you should consider increasing the maximum.
2729 the kernel allocated file handles dynamically,
2730 but it didn't free them again.
2731 Instead the free file handles were kept in a list for reallocation;
2732 the "free file handles" value indicates the size of that list.
2733 A large number of free file handles indicates that there was
2734 a past peak in the usage of open file handles.
2735 Since Linux 2.6, the kernel does deallocate freed file handles,
2736 and the "free file handles" value is always zero.
2738 .IR /proc/sys/fs/inode-max " (only present until Linux 2.2)"
2739 This file contains the maximum number of in-memory inodes.
2740 This value should be 3-4 times larger
2743 since \fIstdin\fP, \fIstdout\fP
2744 and network sockets also need an inode to handle them.
2745 When you regularly run out of inodes, you need to increase this value.
2747 Starting with Linux 2.4,
2748 there is no longer a static limit on the number of inodes,
2749 and this file is removed.
2751 .I /proc/sys/fs/inode-nr
2752 This file contains the first two values from
2755 .I /proc/sys/fs/inode-state
2757 contains seven numbers:
2759 .IR nr_free_inodes ,
2761 and four dummy values (always zero).
2764 is the number of inodes the system has allocated.
2765 .\" This can be slightly more than
2767 .\" because Linux allocates them one page full at a time.
2769 represents the number of free inodes.
2776 and the system needs to prune the inode list instead of allocating more;
2777 since Linux 2.4, this field is a dummy value (always zero).
2779 .IR /proc/sys/fs/inotify " (since Linux 2.6.13)"
2780 This directory contains files
2781 .IR max_queued_events ", " max_user_instances ", and " max_user_watches ,
2782 that can be used to limit the amount of kernel memory consumed by the
2785 For further details, see
2788 .I /proc/sys/fs/lease-break-time
2789 This file specifies the grace period that the kernel grants to a process
2790 holding a file lease
2792 after it has sent a signal to that process notifying it
2793 that another process is waiting to open the file.
2794 If the lease holder does not remove or downgrade the lease within
2795 this grace period, the kernel forcibly breaks the lease.
2797 .I /proc/sys/fs/leases-enable
2798 This file can be used to enable or disable file leases
2800 on a system-wide basis.
2801 If this file contains the value 0, leases are disabled.
2802 A nonzero value enables leases.
2804 .IR /proc/sys/fs/mqueue " (since Linux 2.6.6)"
2805 This directory contains files
2806 .IR msg_max ", " msgsize_max ", and " queues_max ,
2807 controlling the resources used by POSIX message queues.
2812 .IR /proc/sys/fs/overflowgid " and " /proc/sys/fs/overflowuid
2814 allow you to change the value of the fixed UID and GID.
2815 The default is 65534.
2816 Some filesystems support only 16-bit UIDs and GIDs, although in Linux
2817 UIDs and GIDs are 32 bits.
2818 When one of these filesystems is mounted
2819 with writes enabled, any UID or GID that would exceed 65535 is translated
2820 to the overflow value before being written to disk.
2822 .IR /proc/sys/fs/pipe-max-size " (since Linux 2.6.35)"
2823 The value in this file defines an upper limit for raising the capacity
2828 This limit applies only to unprivileged processes.
2829 The default value for this file is 1,048,576.
2830 The value assigned to this file may be rounded upward,
2831 to reflect the value actually employed for a convenient implementation.
2832 To determine the rounded-up value,
2833 display the contents of this file after assigning a value to it.
2834 The minimum value that can be assigned to this file is the system page size.
2836 .IR /proc/sys/fs/protected_hardlinks " (since Linux 3.6)"
2837 .\" commit 800179c9b8a1e796e441674776d11cd4c05d61d7
2838 When the value in this file is 0,
2839 no restrictions are placed on the creation of hard links
2840 (i.e., this is the historical behaviour before Linux 3.6).
2841 When the value in this file is 1,
2842 a hard link can be created to a target file
2843 only if one of the following conditions is true:
2850 The filesystem UID of the process creating the link matches
2851 the owner (UID) of the target file
2853 .BR credentials (7),
2854 a process's filesystem UID is normally the same as its effective UID).
2856 All of the following conditions are true:
2859 the target is a regular file;
2861 the target file does not have its set-user-ID permission bit enabled;
2863 the target file does not have both its set-group-ID and
2864 group-executable permission bits enabled; and
2866 the caller has permission to read and write the target file
2867 (either via the file's permissions mask or because it has
2868 suitable capabilities).
2872 The default value in this file is 0.
2873 Setting the value to 1
2874 prevents a longstanding class of security issues caused by
2875 hard-link-based time-of-check, time-of-use races,
2876 most commonly seen in world-writable directories such as
2878 The common method of exploiting this flaw
2879 is to cross privilege boundaries when following a given hard link
2880 (i.e., a root process follows a hard link created by another user).
2881 Additionally, on systems without separated partitions,
2882 this stops unauthorized users from "pinning" vulnerable set-user-ID and
2883 set-group-ID files against being upgraded by
2884 the administrator, or linking to special files.
2886 .IR /proc/sys/fs/protected_symlinks " (since Linux 3.6)"
2887 .\" commit 800179c9b8a1e796e441674776d11cd4c05d61d7
2888 When the value in this file is 0,
2889 no restrictions are placed on following symbolic links
2890 (i.e., this is the historical behaviour before Linux 3.6).
2891 When the value in this file is 1, symbolic links are followed only
2892 in the following circumstances:
2895 the filesystem UID of the process following the link matches
2896 the owner (UID) of the symbolic link
2898 .BR credentials (7),
2899 a process's filesystem UID is normally the same as its effective UID);
2901 the link is not in a sticky world-writable directory; or
2903 the symbolic link and its parent directory have the same owner (UID)
2906 A system call that fails to follow a symbolic link
2907 because of the above restrictions returns the error
2912 The default value in this file is 0.
2913 Setting the value to 1 avoids a longstanding class of security issues
2914 based on time-of-check, time-of-use races when accessing symbolic links.
2916 .IR /proc/sys/fs/suid_dumpable " (since Linux 2.6.13)"
2917 .\" The following is based on text from Documentation/sysctl/kernel.txt
2918 The value in this file determines whether core dump files are
2919 produced for set-user-ID or otherwise protected/tainted binaries.
2920 Three different integer values can be specified:
2924 This provides the traditional (pre-Linux 2.6.13) behavior.
2925 A core dump will not be produced for a process which has
2926 changed credentials (by calling
2929 or similar, or by executing a set-user-ID or set-group-ID program)
2930 or whose binary does not have read permission enabled.
2933 All processes dump core when possible.
2934 The core dump is owned by the filesystem user ID of the dumping process
2935 and no security is applied.
2936 This is intended for system debugging situations only.
2937 Ptrace is unchecked.
2939 \fI2\ ("suidsafe")\fP
2940 Any binary which normally would not be dumped (see "0" above)
2941 is dumped readable by root only.
2942 This allows the user to remove the core dump file but not to read it.
2943 For security reasons core dumps in this mode will not overwrite one
2944 another or other files.
2945 This mode is appropriate when administrators are
2946 attempting to debug problems in a normal environment.
2948 Additionally, since Linux 3.6,
2949 .\" 9520628e8ceb69fa9a4aee6b57f22675d9e1b709
2950 .I /proc/sys/kernel/core_pattern
2951 must either be an absolute pathname
2952 or a pipe command, as detailed in
2954 Warnings will be written to the kernel log if
2956 does not follow these rules, and no core dump will be produced.
2957 .\" 54b501992dd2a839e94e76aa392c392b55080ce8
2960 .I /proc/sys/fs/super-max
2962 controls the maximum number of superblocks, and
2963 thus the maximum number of mounted filesystems the kernel
2965 You need increase only
2967 if you need to mount more filesystems than the current value in
2971 .I /proc/sys/fs/super-nr
2973 contains the number of filesystems currently mounted.
2976 This directory contains files controlling a range of kernel parameters,
2979 .I /proc/sys/kernel/acct
2981 contains three numbers:
2986 If BSD-style process accounting is enabled, these values control
2988 If free space on filesystem where the log lives goes below
2990 percent, accounting suspends.
2991 If free space gets above
2993 percent, accounting resumes.
2996 how often the kernel checks the amount of free space (value is in
2998 Default values are 4, 2 and 30.
2999 That is, suspend accounting if 2% or less space is free; resume it
3000 if 4% or more space is free; consider information about amount of free space
3001 valid for 30 seconds.
3003 .IR /proc/sys/kernel/cap_last_cap " (since Linux 3.2)"
3005 .BR capabilities (7).
3007 .IR /proc/sys/kernel/cap-bound " (from Linux 2.2 to 2.6.24)"
3008 This file holds the value of the kernel
3009 .I "capability bounding set"
3010 (expressed as a signed decimal number).
3011 This set is ANDed against the capabilities permitted to a process
3014 Starting with Linux 2.6.25,
3015 the system-wide capability bounding set disappeared,
3016 and was replaced by a per-thread bounding set; see
3017 .BR capabilities (7).
3019 .I /proc/sys/kernel/core_pattern
3023 .I /proc/sys/kernel/core_uses_pid
3027 .I /proc/sys/kernel/ctrl-alt-del
3029 controls the handling of Ctrl-Alt-Del from the keyboard.
3030 When the value in this file is 0, Ctrl-Alt-Del is trapped and
3033 program to handle a graceful restart.
3034 When the value is greater than zero, Linux's reaction to a Vulcan
3035 Nerve Pinch (tm) will be an immediate reboot, without even
3036 syncing its dirty buffers.
3037 Note: when a program (like dosemu) has the keyboard in "raw"
3038 mode, the ctrl-alt-del is intercepted by the program before it
3039 ever reaches the kernel tty layer, and it's up to the program
3040 to decide what to do with it.
3042 .IR /proc/sys/kernel/dmesg_restrict " (since Linux 2.6.37)"
3043 The value in this file determines who can see kernel syslog contents.
3044 A value of 0 in this file imposes no restrictions.
3045 If the value is 1, only privileged users can read the kernel syslog.
3050 .\" commit 620f6e8e855d6d447688a5f67a4e176944a084e8
3053 capability may change the value in this file.
3055 .IR /proc/sys/kernel/domainname " and " /proc/sys/kernel/hostname
3056 can be used to set the NIS/YP domainname and the
3057 hostname of your box in exactly the same way as the commands
3065 .RB "#" " echo \(aqdarkstar\(aq > /proc/sys/kernel/hostname"
3066 .RB "#" " echo \(aqmydomain\(aq > /proc/sys/kernel/domainname"
3070 has the same effect as
3074 .RB "#" " hostname \(aqdarkstar\(aq"
3075 .RB "#" " domainname \(aqmydomain\(aq"
3079 Note, however, that the classic darkstar.frop.org has the
3080 hostname "darkstar" and DNS (Internet Domain Name Server)
3081 domainname "frop.org", not to be confused with the NIS (Network
3082 Information Service) or YP (Yellow Pages) domainname.
3084 domain names are in general different.
3085 For a detailed discussion
3090 .I /proc/sys/kernel/hotplug
3092 contains the path for the hotplug policy agent.
3093 The default value in this file is
3096 .I /proc/sys/kernel/htab-reclaim
3097 (PowerPC only) If this file is set to a nonzero value,
3100 .IR Documentation/powerpc/ppc_htab.txt )
3102 each time the system hits the idle loop.
3104 .IR /proc/sys/kernel/kptr_restrict " (since Linux 2.6.38)"
3105 .\" 455cd5ab305c90ffc422dd2e0fb634730942b257
3106 The value in this file determines whether kernel addresses are exposed via
3108 files and other interfaces.
3109 A value of 0 in this file imposes no restrictions.
3110 If the value is 1, kernel pointers printed using the
3112 format specifier will be replaced with zeros unless the user has the
3115 If the value is 2, kernel pointers printed using the
3117 format specifier will be replaced with zeros regardless
3118 of the user's capabilities.
3119 The initial default value for this file was 1,
3120 but the default was changed
3121 .\" commit 411f05f123cbd7f8aa1edcae86970755a6e2a9d9
3122 to 0 in Linux 2.6.39.
3124 .\" commit 620f6e8e855d6d447688a5f67a4e176944a084e8
3127 capability can change the value in this file.
3129 .I /proc/sys/kernel/l2cr
3130 (PowerPC only) This file
3131 contains a flag that controls the L2 cache of G3 processor
3133 If 0, the cache is disabled.
3136 .I /proc/sys/kernel/modprobe
3137 This file contains the path for the kernel module loader.
3138 The default value is
3139 .IR /sbin/modprobe .
3140 The file is present only if the kernel is built with the
3143 in Linux 2.6.26 and earlier)
3145 It is described by the Linux kernel source file
3146 .I Documentation/kmod.txt
3147 (present only in kernel 2.4 and earlier).
3149 .IR /proc/sys/kernel/modules_disabled " (since Linux 2.6.31)"
3150 .\" 3d43321b7015387cfebbe26436d0e9d299162ea1
3151 .\" From Documentation/sysctl/kernel.txt
3152 A toggle value indicating if modules are allowed to be loaded
3153 in an otherwise modular kernel.
3154 This toggle defaults to off (0), but can be set true (1).
3155 Once true, modules can be neither loaded nor unloaded,
3156 and the toggle cannot be set back to false.
3157 The file is present only if the kernel is built with the
3161 .IR /proc/sys/kernel/msgmax " (since Linux 2.2)"
3163 a system-wide limit specifying the maximum number of bytes in
3164 a single message written on a System V message queue.
3166 .IR /proc/sys/kernel/msgmni " (since Linux 2.4)"
3167 This file defines the system-wide limit on the number of
3168 message queue identifiers.
3170 .IR /proc/sys/kernel/msgmnb " (since Linux 2.2)"
3171 This file defines a system-wide parameter used to initialize the
3173 setting for subsequently created message queues.
3176 setting specifies the maximum number of bytes that may be written to the
3179 .IR /proc/sys/kernel/ngroups_max " (since Linux 2.6.4)"
3180 This is a read-only file that displays the upper limit on the
3181 number of a process's group memberships.
3183 .IR /proc/sys/kernel/ostype " and " /proc/sys/kernel/osrelease
3188 .IR /proc/sys/kernel/overflowgid " and " /proc/sys/kernel/overflowuid
3189 These files duplicate the files
3190 .I /proc/sys/fs/overflowgid
3192 .IR /proc/sys/fs/overflowuid .
3194 .I /proc/sys/kernel/panic
3195 This file gives read/write access to the kernel variable
3197 If this is zero, the kernel will loop on a panic; if nonzero,
3198 it indicates that the kernel should autoreboot after this number
3201 software watchdog device driver, the recommended setting is 60.
3203 .IR /proc/sys/kernel/panic_on_oops " (since Linux 2.5.68)"
3204 This file controls the kernel's behavior when an oops
3205 or BUG is encountered.
3206 If this file contains 0, then the system
3207 tries to continue operation.
3208 If it contains 1, then the system
3209 delays a few seconds (to give klogd time to record the oops output)
3212 .I /proc/sys/kernel/panic
3213 file is also nonzero, then the machine will be rebooted.
3215 .IR /proc/sys/kernel/pid_max " (since Linux 2.5.34)"
3216 This file specifies the value at which PIDs wrap around
3217 (i.e., the value in this file is one greater than the maximum PID).
3218 The default value for this file, 32768,
3219 results in the same range of PIDs as on earlier kernels.
3220 On 32-bit platforms, 32768 is the maximum value for
3224 can be set to any value up to 2^22
3225 .RB ( PID_MAX_LIMIT ,
3226 approximately 4 million).
3227 .\" Prior to 2.6.10, pid_max could also be raised above 32768 on 32-bit
3228 .\" platforms, but this broke /proc/[pid]
3229 .\" See http://marc.theaimsgroup.com/?l=linux-kernel&m=109513010926152&w=2
3231 .IR /proc/sys/kernel/powersave-nap " (PowerPC only)"
3232 This file contains a flag.
3233 If set, Linux-PPC will use the "nap" mode of
3235 otherwise the "doze" mode will be used.
3237 .I /proc/sys/kernel/printk
3238 The four values in this file are
3239 .IR console_loglevel ,
3240 .IR default_message_loglevel ,
3241 .IR minimum_console_level ,
3243 .IR default_console_loglevel .
3244 These values influence
3246 behavior when printing or logging error messages.
3249 for more info on the different loglevels.
3250 Messages with a higher priority than
3252 will be printed to the console.
3253 Messages without an explicit priority will be printed with priority
3254 .IR default_message_level .
3255 .I minimum_console_loglevel
3256 is the minimum (highest) value to which
3259 .I default_console_loglevel
3260 is the default value for
3261 .IR console_loglevel .
3263 .IR /proc/sys/kernel/pty " (since Linux 2.6.4)"
3264 This directory contains two files relating to the number of UNIX 98
3265 pseudoterminals (see
3269 .I /proc/sys/kernel/pty/max
3270 This file defines the maximum number of pseudoterminals.
3272 .I /proc/sys/kernel/pty/nr
3274 indicates how many pseudoterminals are currently in use.
3276 .I /proc/sys/kernel/random
3278 contains various parameters controlling the operation of the file
3282 for further information.
3284 .IR /proc/sys/kernel/random/uuid " (since Linux 2.4)"
3285 Each read from this read-only file returns a randomly generated 128-bit UUID,
3286 as a string in the standard UUID format.
3288 .I /proc/sys/kernel/real-root-dev
3289 This file is documented in the Linux kernel source file
3290 .IR Documentation/initrd.txt .
3292 .IR /proc/sys/kernel/reboot-cmd " (Sparc only) "
3293 This file seems to be a way to give an argument to the SPARC
3294 ROM/Flash boot loader.
3295 Maybe to tell it what to do after
3298 .I /proc/sys/kernel/rtsig-max
3299 (Only in kernels up to and including 2.6.7; see
3301 This file can be used to tune the maximum number
3302 of POSIX real-time (queued) signals that can be outstanding
3305 .I /proc/sys/kernel/rtsig-nr
3306 (Only in kernels up to and including 2.6.7.)
3307 This file shows the number POSIX real-time signals currently queued.
3309 .IR /proc/sys/kernel/sched_rr_timeslice_ms " (since Linux 3.9)"
3311 .BR sched_rr_get_interval (2).
3313 .IR /proc/sys/kernel/sem " (since Linux 2.4)"
3314 This file contains 4 numbers defining limits for System V IPC semaphores.
3315 These fields are, in order:
3318 The maximum semaphores per semaphore set.
3320 A system-wide limit on the number of semaphores in all semaphore sets.
3322 The maximum number of operations that may be specified in a
3326 A system-wide limit on the maximum number of semaphore identifiers.
3329 .I /proc/sys/kernel/sg-big-buff
3331 shows the size of the generic SCSI device (sg) buffer.
3332 You can't tune it just yet, but you could change it at
3333 compile time by editing
3334 .I include/scsi/sg.h
3338 However, there shouldn't be any reason to change this value.
3340 .IR /proc/sys/kernel/shm_rmid_forced " (since Linux 3.1)"
3341 .\" commit b34a6b1da371ed8af1221459a18c67970f7e3d53
3342 .\" See also Documentation/sysctl/kernel.txt
3343 If this file is set to 1, all System V shared memory segments will
3344 be marked for destruction as soon as the number of attached processes
3346 in other words, it is no longer possible to create shared memory segments
3347 that exist independently of any attached process.
3349 The effect is as though a
3352 is performed on all existing segments as well as all segments
3353 created in the future (until this file is reset to 0).
3354 Note that existing segments that are attached to no process will be
3355 immediately destroyed when this file is set to 1.
3356 Setting this option will also destroy segments that were created,
3358 upon termination of the process that created the segment with
3361 Setting this file to 1 provides a way of ensuring that
3362 all System V shared memory segments are counted against the
3363 resource usage and resource limits (see the description of
3367 of at least one process.
3369 Because setting this file to 1 produces behavior that is nonstandard
3370 and could also break existing applications,
3371 the default value in this file is 0.
3372 Only set this file to 1 if you have a good understanding
3373 of the semantics of the applications using
3374 System V shared memory on your system.
3376 .IR /proc/sys/kernel/shmall " (since Linux 2.2)"
3378 contains the system-wide limit on the total number of pages of
3379 System V shared memory.
3381 .IR /proc/sys/kernel/shmmax " (since Linux 2.2)"
3383 can be used to query and set the run-time limit
3384 on the maximum (System V IPC) shared memory segment size that can be
3386 Shared memory segments up to 1GB are now supported in the
3388 This value defaults to
3391 .IR /proc/sys/kernel/shmmni " (since Linux 2.4)"
3393 specifies the system-wide maximum number of System V shared memory
3394 segments that can be created.
3396 .I /proc/sys/kernel/sysrq
3397 This file controls the functions allowed to be invoked by the SysRq key.
3399 the file contains 1 meaning that every possible SysRq request is allowed
3400 (in older kernel versions, SysRq was disabled by default,
3401 and you were required to specifically enable it at run-time,
3402 but this is not the case any more).
3403 Possible values in this file are:
3405 0 - disable sysrq completely
3406 1 - enable all functions of sysrq
3407 >1 - bit mask of allowed sysrq functions, as follows:
3408 2 - enable control of console logging level
3409 4 - enable control of keyboard (SAK, unraw)
3410 8 - enable debugging dumps of processes etc.
3411 16 - enable sync command
3412 32 - enable remount read-only
3413 64 - enable signaling of processes (term, kill, oom-kill)
3414 128 - allow reboot/poweroff
3415 256 - allow nicing of all real-time tasks
3417 This file is present only if the
3418 .B CONFIG_MAGIC_SYSRQ
3419 kernel configuration option is enabled.
3420 For further details see the Linux kernel source file
3421 .IR Documentation/sysrq.txt .
3423 .I /proc/sys/kernel/version
3424 This file contains a string like:
3426 #5 Wed Feb 25 21:49:24 MET 1998
3429 this is the fifth kernel built from this source base and the
3430 date behind it indicates the time the kernel was built.
3432 .IR /proc/sys/kernel/threads-max " (since Linux 2.3.11)"
3433 This file specifies the system-wide limit on the number of
3434 threads (tasks) that can be created on the system.
3436 .IR /proc/sys/kernel/zero-paged " (PowerPC only) "
3439 When enabled (nonzero), Linux-PPC will pre-zero pages in
3440 the idle loop, possibly speeding up get_free_pages.
3443 This directory contains networking stuff.
3444 Explanations for some of the files under this directory can be found in
3449 .I /proc/sys/net/core/somaxconn
3450 This file defines a ceiling value for the
3456 manual page for details.
3459 This directory may be empty.
3462 This directory supports Sun remote procedure call for network filesystem
3464 On some systems, it is not present.
3467 This directory contains files for memory management tuning, buffer and
3470 .IR /proc/sys/vm/drop_caches " (since Linux 2.6.16)"
3471 Writing to this file causes the kernel to drop clean caches, dentries, and
3472 inodes from memory, causing that memory to become free.
3473 This can be useful for memory management testing and
3474 performing reproducible filesystem benchmarks.
3475 Because writing to this file causes the benefits of caching to be lost,
3476 it can degrade overall system performance.
3478 To free pagecache, use:
3480 echo 1 > /proc/sys/vm/drop_caches
3482 To free dentries and inodes, use:
3484 echo 2 > /proc/sys/vm/drop_caches
3486 To free pagecache, dentries and inodes, use:
3488 echo 3 > /proc/sys/vm/drop_caches
3490 Because writing to this file is a nondestructive operation and dirty objects
3491 are not freeable, the
3496 .IR /proc/sys/vm/legacy_va_layout " (since Linux 2.6.9)"
3497 .\" The following is from Documentation/filesystems/proc.txt
3498 If nonzero, this disables the new 32-bit memory-mapping layout;
3499 the kernel will use the legacy (2.4) layout for all processes.
3501 .IR /proc/sys/vm/memory_failure_early_kill " (since Linux 2.6.32)"
3502 .\" The following is based on the text in Documentation/sysctl/vm.txt
3503 Control how to kill processes when an uncorrected memory error
3504 (typically a 2-bit error in a memory module)
3505 that cannot be handled by the kernel
3506 is detected in the background by hardware.
3507 In some cases (like the page still having a valid copy on disk),
3508 the kernel will handle the failure
3509 transparently without affecting any applications.
3510 But if there is no other up-to-date copy of the data,
3511 it will kill processes to prevent any data corruptions from propagating.
3513 The file has one of the following values:
3516 Kill all processes that have the corrupted-and-not-reloadable page mapped
3517 as soon as the corruption is detected.
3518 Note this is not supported for a few types of pages, like kernel internally
3519 allocated data or the swap cache, but works for the majority of user pages.
3521 Only unmap the corrupted page from all processes and kill only a process
3522 that tries to access it.
3525 The kill is performed using a
3531 Processes can handle this if they want to; see
3535 This feature is active only on architectures/platforms with advanced machine
3536 check handling and depends on the hardware capabilities.
3538 Applications can override the
3539 .I memory_failure_early_kill
3540 setting individually with the
3545 Only present if the kernel was configured with
3546 .BR CONFIG_MEMORY_FAILURE .
3548 .IR /proc/sys/vm/memory_failure_recovery " (since Linux 2.6.32)"
3549 .\" The following is based on the text in Documentation/sysctl/vm.txt
3550 Enable memory failure recovery (when supported by the platform)
3555 Always panic on a memory failure.
3558 Only present if the kernel was configured with
3559 .BR CONFIG_MEMORY_FAILURE .
3561 .IR /proc/sys/vm/oom_dump_tasks " (since Linux 2.6.25)"
3562 .\" The following is from Documentation/sysctl/vm.txt
3563 Enables a system-wide task dump (excluding kernel threads) to be
3564 produced when the kernel performs an OOM-killing.
3565 The dump includes the following information
3566 for each task (thread, process):
3567 thread ID, real user ID, thread group ID (process ID),
3568 virtual memory size, resident set size,
3569 the CPU that the task is scheduled on,
3570 oom_adj score (see the description of
3571 .IR /proc/[pid]/oom_adj ),
3573 This is helpful to determine why the OOM-killer was invoked
3574 and to identify the rogue task that caused it.
3576 If this contains the value zero, this information is suppressed.
3577 On very large systems with thousands of tasks,
3578 it may not be feasible to dump the memory state information for each one.
3579 Such systems should not be forced to incur a performance penalty in
3580 OOM situations when the information may not be desired.
3582 If this is set to nonzero, this information is shown whenever the
3583 OOM-killer actually kills a memory-hogging task.
3585 The default value is 0.
3587 .IR /proc/sys/vm/oom_kill_allocating_task " (since Linux 2.6.24)"
3588 .\" The following is from Documentation/sysctl/vm.txt
3589 This enables or disables killing the OOM-triggering task in
3590 out-of-memory situations.
3592 If this is set to zero, the OOM-killer will scan through the entire
3593 tasklist and select a task based on heuristics to kill.
3594 This normally selects a rogue memory-hogging task that
3595 frees up a large amount of memory when killed.
3597 If this is set to nonzero, the OOM-killer simply kills the task that
3598 triggered the out-of-memory condition.
3599 This avoids a possibly expensive tasklist scan.
3602 .I /proc/sys/vm/panic_on_oom
3603 is nonzero, it takes precedence over whatever value is used in
3604 .IR /proc/sys/vm/oom_kill_allocating_task .
3606 The default value is 0.
3608 .I /proc/sys/vm/overcommit_memory
3609 This file contains the kernel virtual memory accounting mode.
3613 0: heuristic overcommit (this is the default)
3615 1: always overcommit, never check
3617 2: always check, never overcommit
3624 are not checked, and the default check is very weak,
3625 leading to the risk of getting a process "OOM-killed".
3626 Under Linux 2.4 any nonzero value implies mode 1.
3627 In mode 2 (available since Linux 2.6), the total virtual address space
3628 on the system is limited to (SS + RAM*(r/100)),
3629 where SS is the size of the swap space, and RAM
3630 is the size of the physical memory, and r is the contents of the file
3631 .IR /proc/sys/vm/overcommit_ratio .
3633 .I /proc/sys/vm/overcommit_ratio
3634 See the description of
3635 .IR /proc/sys/vm/overcommit_memory .
3637 .IR /proc/sys/vm/panic_on_oom " (since Linux 2.6.18)"
3638 .\" The following is adapted from Documentation/sysctl/vm.txt
3639 This enables or disables a kernel panic in
3640 an out-of-memory situation.
3642 If this file is set to the value 0,
3643 the kernel's OOM-killer will kill some rogue process.
3644 Usually, the OOM-killer is able to kill a rogue process and the
3645 system will survive.
3647 If this file is set to the value 1,
3648 then the kernel normally panics when out-of-memory happens.
3649 However, if a process limits allocations to certain nodes
3650 using memory policies
3655 and those nodes reach memory exhaustion status,
3656 one process may be killed by the OOM-killer.
3657 No panic occurs in this case:
3658 because other nodes' memory may be free,
3659 this means the system as a whole may not have reached
3660 an out-of-memory situation yet.
3662 If this file is set to the value 2,
3663 the kernel always panics when an out-of-memory condition occurs.
3665 The default value is 0.
3666 1 and 2 are for failover of clustering.
3667 Select either according to your policy of failover.
3669 .IR /proc/sys/vm/swappiness
3670 .\" The following is from Documentation/sysctl/vm.txt
3671 The value in this file controls how aggressively the kernel will swap
3673 Higher values increase aggressiveness, lower values
3674 decrease aggressiveness.
3675 The default value is 60.
3677 .IR /proc/sysrq-trigger " (since Linux 2.4.21)"
3678 Writing a character to this file triggers the same SysRq function as
3679 typing ALT-SysRq-<character> (see the description of
3680 .IR /proc/sys/kernel/sysrq ).
3681 This file is normally writable only by
3683 For further details see the Linux kernel source file
3684 .IR Documentation/sysrq.txt .
3687 Subdirectory containing the pseudo-files
3688 .IR msg ", " sem " and " shm "."
3689 These files list the System V Interprocess Communication (IPC) objects
3690 (respectively: message queues, semaphores, and shared memory)
3691 that currently exist on the system,
3692 providing similar information to that available via
3694 These files have headers and are formatted (one IPC object per line)
3695 for easy understanding.
3697 provides further background on the information shown by these files.
3700 Subdirectory containing the pseudo-files and subdirectories for
3701 tty drivers and line disciplines.
3704 This file contains two numbers: the uptime of the system (seconds),
3705 and the amount of time spent in idle process (seconds).
3708 This string identifies the kernel version that is currently running.
3709 It includes the contents of
3710 .IR /proc/sys/kernel/ostype ,
3711 .I /proc/sys/kernel/osrelease
3713 .IR /proc/sys/kernel/version .
3718 Linux version 1.0.9 (quinlan@phaze) #1 Sat May 14 01:51:54 EDT 1994
3722 .\" FIXME Document /proc/timer_list
3724 .\" .IR /proc/timer_list " (since Linux 2.6.21)"
3725 .\" See the 2.6.21 Change log
3726 .\" FIXME Document /proc/timer_stats
3728 .\" .IR /proc/timer_stats " (since Linux 2.6.21)"
3729 .\" See the 2.6.21 Change log
3731 .IR /proc/vmstat " (since Linux 2.6)"
3732 This file displays various virtual memory statistics.
3734 .IR /proc/zoneinfo " (since Linux 2.6.13)"
3735 This file display information about memory zones.
3736 This is useful for analyzing virtual memory behavior.
3737 .\" FIXME more should be said about /proc/zoneinfo
3739 Many strings (i.e., the environment and command line) are in
3740 the internal format, with subfields terminated by null bytes (\(aq\\0\(aq),
3742 may find that things are more readable if you use \fIod \-c\fP or \fItr
3743 "\\000" "\\n"\fP to read them.
3744 Alternatively, \fIecho \`cat <file>\`\fP works well.
3746 This manual page is incomplete, possibly inaccurate, and is the kind
3747 of thing that needs to be updated very often.
3748 .\" .SH ACKNOWLEDGEMENTS
3749 .\" The material on /proc/sys/fs and /proc/sys/kernel is closely based on
3750 .\" kernel source documentation files written by Rik van Riel.
3778 The Linux kernel source files:
3779 .IR Documentation/filesystems/proc.txt
3780 .IR Documentation/sysctl/fs.txt ,
3781 .IR Documentation/sysctl/kernel.txt ,
3782 .IR Documentation/sysctl/net.txt ,
3784 .IR Documentation/sysctl/vm.txt .
3786 This page is part of release 3.65 of the Linux
3789 A description of the project,
3790 and information about reporting bugs,
3792 \%http://www.kernel.org/doc/man\-pages/.