1PROC(5) Linux Programmer's Manual PROC(5)
2
6 proc - process information pseudo-file system
7
9 The proc file system is a pseudo-file system which is used as an inter‐
10 face to kernel data structures. It is commonly mounted at /proc. Most
11 of it is read-only, but some files allow kernel variables to be
12 changed.
13 The following outline gives a quick tour through the /proc hierarchy.
15 /proc/[pid]
17 There is a numerical subdirectory for each running process; the
18 subdirectory is named by the process ID. Each such subdirectory
19 contains the following pseudo-files and directories.
20 /proc/[pid]/auxv (since 2.6.0-test7)
22 This contains the contents of the ELF interpreter information
23 passed to the process at exec time. The format is one unsigned
24 long ID plus one unsigned long value for each entry. The last
25 entry contains two zeros.
26 /proc/[pid]/cgroup (since Linux 2.6.24)
28 This file describes control groups to which the process/task
29 belongs. For each cgroup hierarchy there is one entry contain‐
30 ing colon-separated fields of the form:
31 5:cpuacct,cpu,cpuset:/daemons
33 The colon-separated fields are, from left to right:
35 1. hierarchy ID number
37 2. set of subsystems bound to the hierarchy
39 3. control group in the hierarchy to which the process
41 belongs
42 This file is present only if the CONFIG_CGROUPS kernel configu‐
44 ration option is enabled.
45 /proc/[pid]/cmdline
47 This holds the complete command line for the process, unless the
48 process is a zombie. In the latter case, there is nothing in
49 this file: that is, a read on this file will return 0 charac‐
50 ters. The command-line arguments appear in this file as a set
51 of strings separated by null bytes ('\0'), with a further null
52 byte after the last string.
53 /proc/[pid]/coredump_filter (since kernel 2.6.23)
55 See core(5).
56 /proc/[pid]/cpuset (since kernel 2.6.12)
58 See cpuset(7).
59 /proc/[pid]/cwd
61 This is a symbolic link to the current working directory of the
62 process. To find out the current working directory of process
63 20, for instance, you can do this:
64 $ cd /proc/20/cwd; /bin/pwd
66 Note that the pwd command is often a shell built-in, and might
68 not work properly. In bash(1), you may use pwd -P.
69 In a multithreaded process, the contents of this symbolic link
71 are not available if the main thread has already terminated
72 (typically by calling pthread_exit(3)).
73 /proc/[pid]/environ
75 This file contains the environment for the process. The entries
76 are separated by null bytes ('\0'), and there may be a null byte
77 at the end. Thus, to print out the environment of process 1,
78 you would do:
79 $ strings /proc/1/environ
81 /proc/[pid]/exe
83 Under Linux 2.2 and later, this file is a symbolic link contain‐
84 ing the actual pathname of the executed command. This symbolic
85 link can be dereferenced normally; attempting to open it will
86 open the executable. You can even type /proc/[pid]/exe to run
87 another copy of the same executable as is being run by process
88 [pid]. In a multithreaded process, the contents of this sym‐
89 bolic link are not available if the main thread has already ter‐
90 minated (typically by calling pthread_exit(3)).
91 Under Linux 2.0 and earlier /proc/[pid]/exe is a pointer to the
93 binary which was executed, and appears as a symbolic link. A
94 readlink(2) call on this file under Linux 2.0 returns a string
95 in the format:
96 [device]:inode
98 For example, [0301]:1502 would be inode 1502 on device major 03
100 (IDE, MFM, etc. drives) minor 01 (first partition on the first
101 drive).
102 find(1) with the -inum option can be used to locate the file.
104 /proc/[pid]/fd/
106 This is a subdirectory containing one entry for each file which
107 the process has open, named by its file descriptor, and which is
108 a symbolic link to the actual file. Thus, 0 is standard input,
109 1 standard output, 2 standard error, etc.
110 For file descriptors for pipes and sockets, the entries will be
112 symbolic links whose content is the file type with the inode. A
113 readlink(2) call on this file returns a string in the format:
114 type:[inode]
116 For example, socket:[2248868] will be a socket and its inode is
118 2248868. For sockets, that inode can be used to find more
119 information in one of the files under /proc/net/.
120 For file descriptors that have no corresponding inode (e.g.,
122 file descriptors produced by epoll_create(2), eventfd(2), ino‐
123 tify_init(2), signalfd(2), and timerfd(2)), the entry will be a
124 symbolic link with contents of the form
125 anon_inode:<file-type>
127 In some cases, the file-type is surrounded by square brackets.
129 For example, an epoll file descriptor will have a symbolic link
131 whose content is the string anon_inode:[eventpoll].
132 In a multithreaded process, the contents of this directory are
134 not available if the main thread has already terminated (typi‐
135 cally by calling pthread_exit(3)).
136 Programs that will take a filename as a command-line argument,
138 but will not take input from standard input if no argument is
139 supplied, or that write to a file named as a command-line argu‐
140 ment, but will not send their output to standard output if no
141 argument is supplied, can nevertheless be made to use standard
142 input or standard out using /proc/[pid]/fd. For example, assum‐
143 ing that -i is the flag designating an input file and -o is the
144 flag designating an output file:
145 $ foobar -i /proc/self/fd/0 -o /proc/self/fd/1 ...
147 and you have a working filter.
149 /proc/self/fd/N is approximately the same as /dev/fd/N in some
151 UNIX and UNIX-like systems. Most Linux MAKEDEV scripts symboli‐
152 cally link /dev/fd to /proc/self/fd, in fact.
153 Most systems provide symbolic links /dev/stdin, /dev/stdout, and
155 /dev/stderr, which respectively link to the files 0, 1, and 2 in
156 /proc/self/fd. Thus the example command above could be written
157 as:
158 $ foobar -i /dev/stdin -o /dev/stdout ...
160 /proc/[pid]/fdinfo/ (since kernel 2.6.22)
162 This is a subdirectory containing one entry for each file which
163 the process has open, named by its file descriptor. The con‐
164 tents of each file can be read to obtain information about the
165 corresponding file descriptor, for example:
166 $ cat /proc/12015/fdinfo/4
168 pos: 1000
169 flags: 01002002
170 The pos field is a decimal number showing the current file off‐
172 set. The flags field is an octal number that displays the file
173 access mode and file status flags (see open(2)).
174 The files in this directory are readable only by the owner of
176 the process.
177 /proc/[pid]/io (since kernel 2.6.20)
179 This file contains I/O statistics for the process, for example:
180 # cat /proc/3828/io
182 rchar: 323934931
183 wchar: 323929600
184 syscr: 632687
185 syscw: 632675
186 read_bytes: 0
187 write_bytes: 323932160
188 cancelled_write_bytes: 0
189 The fields are as follows:
191 rchar: characters read
193 The number of bytes which this task has caused to be read
194 from storage. This is simply the sum of bytes which this
195 process passed to read(2) and similar system calls. It
196 includes things such as terminal I/O and is unaffected by
197 whether or not actual physical disk I/O was required (the
198 read might have been satisfied from pagecache).
199 wchar: characters written
201 The number of bytes which this task has caused, or shall
202 cause to be written to disk. Similar caveats apply here
203 as with rchar.
204 syscr: read syscalls
206 Attempt to count the number of read I/O operations—that
207 is, system calls such as read(2) and pread(2).
208 syscw: write syscalls
210 Attempt to count the number of write I/O operations—that
211 is, system calls such as write(2) and pwrite(2).
212 read_bytes: bytes read
214 Attempt to count the number of bytes which this process
215 really did cause to be fetched from the storage layer.
216 This is accurate for block-backed filesystems.
217 write_bytes: bytes written
219 Attempt to count the number of bytes which this process
220 caused to be sent to the storage layer.
221 cancelled_write_bytes:
223 The big inaccuracy here is truncate. If a process writes
224 1MB to a file and then deletes the file, it will in fact
225 perform no writeout. But it will have been accounted as
226 having caused 1MB of write. In other words: this field
227 represents the number of bytes which this process caused
228 to not happen, by truncating pagecache. A task can cause
229 "negative" I/O too. If this task truncates some dirty
230 pagecache, some I/O which another task has been accounted
231 for (in its write_bytes) will not be happening.
232 Note: In the current implementation, things are a bit racy on
234 32-bit systems: if process A reads process B's /proc/[pid]/io
235 while process B is updating one of these 64-bit counters,
236 process A could see an intermediate result.
237 /proc/[pid]/limits (since kernel 2.6.24)
239 This file displays the soft limit, hard limit, and units of mea‐
240 surement for each of the process's resource limits (see getr‐
241 limit(2)). Up to and including Linux 2.6.35, this file is pro‐
242 tected to allow reading only by the real UID of the process.
243 Since Linux 2.6.36, this file is readable by all users on the
244 system.
245 /proc/[pid]/map_files/ (since kernel 3.3)
247 This subdirectory contains entries corresponding to memory-
248 mapped files (see mmap(2)). Entries are named by memory region
249 start and end address pair (expressed as hexadecimal numbers),
250 and are symbolic links to the mapped files themselves. Here is
251 an example, with the output wrapped and reformatted to fit on an
252 80-column display:
253 $ ls -l /proc/self/map_files/
255 lr--------. 1 root root 64 Apr 16 21:31
256 3252e00000-3252e20000 -> /usr/lib64/ld-2.15.so
257 ...
258 Although these entries are present for memory regions that were
260 mapped with the MAP_FILE flag, the way anonymous shared memory
261 (regions created with the MAP_ANON | MAP_SHARED flags) is imple‐
262 mented in Linux means that such regions also appear on this
263 directory. Here is an example where the target file is the
264 deleted /dev/zero one:
265 lrw-------. 1 root root 64 Apr 16 21:33
268 7fc075d2f000-7fc075e6f000 -> /dev/zero (deleted)
269 This directory appears only if the CONFIG_CHECKPOINT_RESTORE
271 kernel configuration option is enabled.
272 /proc/[pid]/maps
274 A file containing the currently mapped memory regions and their
275 access permissions. See mmap(2) for some further information
276 about memory mappings.
277 The format of the file is:
279 address perms offset dev inode pathname
281 00400000-00452000 r-xp 00000000 08:02 173521 /usr/bin/dbus-daemon
282 00651000-00652000 r--p 00051000 08:02 173521 /usr/bin/dbus-daemon
283 00652000-00655000 rw-p 00052000 08:02 173521 /usr/bin/dbus-daemon
284 00e03000-00e24000 rw-p 00000000 00:00 0 [heap]
285 00e24000-011f7000 rw-p 00000000 00:00 0 [heap]
286 ...
287 35b1800000-35b1820000 r-xp 00000000 08:02 135522 /usr/lib64/ld-2.15.so
288 35b1a1f000-35b1a20000 r--p 0001f000 08:02 135522 /usr/lib64/ld-2.15.so
289 35b1a20000-35b1a21000 rw-p 00020000 08:02 135522 /usr/lib64/ld-2.15.so
290 35b1a21000-35b1a22000 rw-p 00000000 00:00 0
291 35b1c00000-35b1dac000 r-xp 00000000 08:02 135870 /usr/lib64/libc-2.15.so
292 35b1dac000-35b1fac000 ---p 001ac000 08:02 135870 /usr/lib64/libc-2.15.so
293 35b1fac000-35b1fb0000 r--p 001ac000 08:02 135870 /usr/lib64/libc-2.15.so
294 35b1fb0000-35b1fb2000 rw-p 001b0000 08:02 135870 /usr/lib64/libc-2.15.so
295 ...
296 f2c6ff8c000-7f2c7078c000 rw-p 00000000 00:00 0 [stack:986]
297 ...
298 7fffb2c0d000-7fffb2c2e000 rw-p 00000000 00:00 0 [stack]
299 7fffb2d48000-7fffb2d49000 r-xp 00000000 00:00 0 [vdso]
300 The address field is the address space in the process that the
302 mapping occupies. The perms field is a set of permissions:
303 r = read
305 w = write
306 x = execute
307 s = shared
308 p = private (copy on write)
309 The offset field is the offset into the file/whatever; dev is
311 the device (major:minor); inode is the inode on that device. 0
312 indicates that no inode is associated with the memory region, as
313 would be the case with BSS (uninitialized data).
314 The pathname field will usually be the file that is backing the
316 mapping. For ELF files, you can easily coordinate with the off‐
317 set field by looking at the Offset field in the ELF program
318 headers (readelf -l).
319 There are additional helpful pseudo-paths:
321 [stack]
323 The initial process's (also known as the main
324 thread's) stack.
325 [stack:<tid>] (since Linux 3.4)
327 A thread's stack (where the <tid> is a thread ID).
328 It corresponds to the /proc/[pid]/task/[tid]/ path.
329 [vdso] The virtual dynamically linked shared object.
331 [heap] The process's heap.
333 If the pathname field is blank, this is an anonymous mapping as
335 obtained via the mmap(2) function. There is no easy way to
336 coordinate this back to a process's source, short of running it
337 through gdb(1), strace(1), or similar.
338 Under Linux 2.0 there is no field giving pathname.
340 /proc/[pid]/mem
342 This file can be used to access the pages of a process's memory
343 through open(2), read(2), and lseek(2).
344 /proc/[pid]/mountinfo (since Linux 2.6.26)
346 This file contains information about mount points. It contains
347 lines of the form:
348 36 35 98:0 /mnt1 /mnt2 rw,noatime master:1 - ext3 /dev/root rw,errors=continue
350 (1)(2)(3) (4) (5) (6) (7) (8) (9) (10) (11)
351 The numbers in parentheses are labels for the descriptions
353 below:
354 (1) mount ID: unique identifier of the mount (may be reused
356 after umount(2)).
357 (2) parent ID: ID of parent mount (or of self for the top of
359 the mount tree).
360 (3) major:minor: value of st_dev for files on file system (see
362 stat(2)).
363 (4) root: root of the mount within the file system.
365 (5) mount point: mount point relative to the process's root.
367 (6) mount options: per-mount options.
369 (7) optional fields: zero or more fields of the form
371 "tag[:value]".
372 (8) separator: marks the end of the optional fields.
374 (9) file system type: name of file system in the form
376 "type[.subtype]".
377 (10) mount source: file system-specific information or "none".
379 (11) super options: per-super block options.
381 Parsers should ignore all unrecognized optional fields. Cur‐
383 rently the possible optional fields are:
384 shared:X mount is shared in peer group X
386 master:X mount is slave to peer group X
388 propagate_from:X mount is slave and receives propagation
390 from peer group X (*)
391 unbindable mount is unbindable
393 (*) X is the closest dominant peer group under the process's
395 root. If X is the immediate master of the mount, or if there is
396 no dominant peer group under the same root, then only the "mas‐
397 ter:X" field is present and not the "propagate_from:X" field.
398 For more information on mount propagation see: Documenta‐
400 tion/filesystems/sharedsubtree.txt in the Linux kernel source
401 tree.
402 /proc/[pid]/mounts (since Linux 2.4.19)
404 This is a list of all the file systems currently mounted in the
405 process's mount namespace. The format of this file is docu‐
406 mented in fstab(5). Since kernel version 2.6.15, this file is
407 pollable: after opening the file for reading, a change in this
408 file (i.e., a file system mount or unmount) causes select(2) to
409 mark the file descriptor as readable, and poll(2) and
410 epoll_wait(2) mark the file as having an error condition.
411 /proc/[pid]/mountstats (since Linux 2.6.17)
413 This file exports information (statistics, configuration infor‐
414 mation) about the mount points in the process's name space.
415 Lines in this file have the form:
416 device /dev/sda7 mounted on /home with fstype ext3 [statistics]
418 ( 1 ) ( 2 ) (3 ) (4)
419 The fields in each line are:
421 (1) The name of the mounted device (or "nodevice" if there is
423 no corresponding device).
424 (2) The mount point within the file system tree.
426 (3) The file system type.
428 (4) Optional statistics and configuration information. Cur‐
430 rently (as at Linux 2.6.26), only NFS file systems export
431 information via this field.
432 This file is readable only by the owner of the process.
434 /proc/[pid]/ns/ (since Linux 3.0)
436 This is a subdirectory containing one entry for each namespace
437 that supports being manipulated by setns(2). For information
438 about namespaces, see clone(2).
439 /proc/[pid]/ns/ipc (since Linux 3.0)
441 Bind mounting this file (see mount(2)) to somewhere else in the
442 filesystem keeps the IPC namespace of the process specified by
443 pid alive even if all processes currently in the namespace ter‐
444 minate.
445 Opening this file returns a file handle for the IPC namespace of
447 the process specified by pid. As long as this file descriptor
448 remains open, the IPC namespace will remain alive, even if all
449 processes in the namespace terminate. The file descriptor can
450 be passed to setns(2).
451 /proc/[pid]/ns/net (since Linux 3.0)
453 Bind mounting this file (see mount(2)) to somewhere else in the
454 filesystem keeps the network namespace of the process specified
455 by pid alive even if all processes in the namespace terminate.
456 Opening this file returns a file handle for the network names‐
458 pace of the process specified by pid. As long as this file
459 descriptor remains open, the network namespace will remain
460 alive, even if all processes in the namespace terminate. The
461 file descriptor can be passed to setns(2).
462 /proc/[pid]/ns/uts (since Linux 3.0)
464 Bind mounting this file (see mount(2)) to somewhere else in the
465 filesystem keeps the UTS namespace of the process specified by
466 pid alive even if all processes currently in the namespace ter‐
467 minate.
468 Opening this file returns a file handle for the UTS namespace of
470 the process specified by pid. As long as this file descriptor
471 remains open, the UTS namespace will remain alive, even if all
472 processes in the namespace terminate. The file descriptor can
473 be passed to setns(2).
474 /proc/[pid]/numa_maps (since Linux 2.6.14)
476 See numa(7).
477 /proc/[pid]/oom_adj (since Linux 2.6.11)
479 This file can be used to adjust the score used to select which
480 process should be killed in an out-of-memory (OOM) situation.
481 The kernel uses this value for a bit-shift operation of the
482 process's oom_score value: valid values are in the range -16 to
483 +15, plus the special value -17, which disables OOM-killing
484 altogether for this process. A positive score increases the
485 likelihood of this process being killed by the OOM-killer; a
486 negative score decreases the likelihood.
487 The default value for this file is 0; a new process inherits its
489 parent's oom_adj setting. A process must be privileged
490 (CAP_SYS_RESOURCE) to update this file.
491 Since Linux 2.6.36, use of this file is deprecated in favor of
493 /proc/[pid]/oom_score_adj.
494 /proc/[pid]/oom_score (since Linux 2.6.11)
496 This file displays the current score that the kernel gives to
497 this process for the purpose of selecting a process for the OOM-
498 killer. A higher score means that the process is more likely to
499 be selected by the OOM-killer. The basis for this score is the
500 amount of memory used by the process, with increases (+) or
501 decreases (-) for factors including:
502 * whether the process creates a lot of children using fork(2)
504 (+);
505 * whether the process has been running a long time, or has used
507 a lot of CPU time (-);
508 * whether the process has a low nice value (i.e., > 0) (+);
510 * whether the process is privileged (-); and
512 * whether the process is making direct hardware access (-).
514 The oom_score also reflects the adjustment specified by the
516 oom_score_adj or oom_adj setting for the process.
517 /proc/[pid]/oom_score_adj (since Linux 2.6.36)
519 This file can be used to adjust the badness heuristic used to
520 select which process gets killed in out-of-memory conditions.
521 The badness heuristic assigns a value to each candidate task
523 ranging from 0 (never kill) to 1000 (always kill) to determine
524 which process is targeted. The units are roughly a proportion
525 along that range of allowed memory the process may allocate
526 from, based on an estimation of its current memory and swap use.
527 For example, if a task is using all allowed memory, its badness
528 score will be 1000. If it is using half of its allowed memory,
529 its score will be 500.
530 There is an additional factor included in the badness score:
532 root processes are given 3% extra memory over other tasks.
533 The amount of "allowed" memory depends on the context in which
535 the OOM-killer was called. If it is due to the memory assigned
536 to the allocating task's cpuset being exhausted, the allowed
537 memory represents the set of mems assigned to that cpuset (see
538 cpuset(7)). If it is due to a mempolicy's node(s) being
539 exhausted, the allowed memory represents the set of mempolicy
540 nodes. If it is due to a memory limit (or swap limit) being
541 reached, the allowed memory is that configured limit. Finally,
542 if it is due to the entire system being out of memory, the
543 allowed memory represents all allocatable resources.
544 The value of oom_score_adj is added to the badness score before
546 it is used to determine which task to kill. Acceptable values
547 range from -1000 (OOM_SCORE_ADJ_MIN) to +1000
548 (OOM_SCORE_ADJ_MAX). This allows user space to control the
549 preference for OOM-killing, ranging from always preferring a
550 certain task or completely disabling it from OOM-killing. The
551 lowest possible value, -1000, is equivalent to disabling OOM-
552 killing entirely for that task, since it will always report a
553 badness score of 0.
554 Consequently, it is very simple for user space to define the
556 amount of memory to consider for each task. Setting a
557 oom_score_adj value of +500, for example, is roughly equivalent
558 to allowing the remainder of tasks sharing the same system,
559 cpuset, mempolicy, or memory controller resources to use at
560 least 50% more memory. A value of -500, on the other hand,
561 would be roughly equivalent to discounting 50% of the task's
562 allowed memory from being considered as scoring against the
563 task.
564 For backward compatibility with previous kernels,
566 /proc/[pid]/oom_adj can still be used to tune the badness score.
567 Its value is scaled linearly with oom_score_adj.
568 Writing to /proc/[pid]/oom_score_adj or /proc/[pid]/oom_adj will
570 change the other with its scaled value.
571 /proc/[pid]/root
573 UNIX and Linux support the idea of a per-process root of the
574 file system, set by the chroot(2) system call. This file is a
575 symbolic link that points to the process's root directory, and
576 behaves as exe, fd/*, etc. do.
577 In a multithreaded process, the contents of this symbolic link
579 are not available if the main thread has already terminated
580 (typically by calling pthread_exit(3)).
581 /proc/[pid]/smaps (since Linux 2.6.14)
583 This file shows memory consumption for each of the process's
584 mappings. For each of mappings there is a series of lines such
585 as the following:
586 08048000-080bc000 r-xp 00000000 03:02 13130 /bin/bash
588 Size: 464 kB
589 Rss: 424 kB
590 Shared_Clean: 424 kB
591 Shared_Dirty: 0 kB
592 Private_Clean: 0 kB
593 Private_Dirty: 0 kB
594 The first of these lines shows the same information as is dis‐
596 played for the mapping in /proc/[pid]/maps. The remaining lines
597 show the size of the mapping, the amount of the mapping that is
598 currently resident in RAM, the number of clean and dirty shared
599 pages in the mapping, and the number of clean and dirty private
600 pages in the mapping.
601 This file is present only if the CONFIG_MMU kernel configuration
603 option is enabled.
604 /proc/[pid]/stat
606 Status information about the process. This is used by ps(1).
607 It is defined in /usr/src/linux/fs/proc/array.c.
608 The fields, in order, with their proper scanf(3) format speci‐
610 fiers, are:
611 pid %d (1) The process ID.
613 comm %s (2) The filename of the executable, in parentheses.
615 This is visible whether or not the executable is
616 swapped out.
617 state %c (3) One character from the string "RSDZTW" where R
619 is running, S is sleeping in an interruptible wait,
620 D is waiting in uninterruptible disk sleep, Z is
621 zombie, T is traced or stopped (on a signal), and W
622 is paging.
623 ppid %d (4) The PID of the parent.
625 pgrp %d (5) The process group ID of the process.
627 session %d (6) The session ID of the process.
629 tty_nr %d (7) The controlling terminal of the process. (The
631 minor device number is contained in the combination
632 of bits 31 to 20 and 7 to 0; the major device number
633 is in bits 15 to 8.)
634 tpgid %d (8) The ID of the foreground process group of the
636 controlling terminal of the process.
637 flags %u (%lu before Linux 2.6.22)
639 (9) The kernel flags word of the process. For bit
640 meanings, see the PF_* defines in the Linux kernel
641 source file include/linux/sched.h. Details depend
642 on the kernel version.
643 minflt %lu (10) The number of minor faults the process has made
645 which have not required loading a memory page from
646 disk.
647 cminflt %lu (11) The number of minor faults that the process's
649 waited-for children have made.
650 majflt %lu (12) The number of major faults the process has made
652 which have required loading a memory page from disk.
653 cmajflt %lu (13) The number of major faults that the process's
655 waited-for children have made.
656 utime %lu (14) Amount of time that this process has been
658 scheduled in user mode, measured in clock ticks
659 (divide by sysconf(_SC_CLK_TCK)). This includes
660 guest time, guest_time (time spent running a virtual
661 CPU, see below), so that applications that are not
662 aware of the guest time field do not lose that time
663 from their calculations.
664 stime %lu (15) Amount of time that this process has been
666 scheduled in kernel mode, measured in clock ticks
667 (divide by sysconf(_SC_CLK_TCK)).
668 cutime %ld (16) Amount of time that this process's waited-for
670 children have been scheduled in user mode, measured
671 in clock ticks (divide by sysconf(_SC_CLK_TCK)).
672 (See also times(2).) This includes guest time,
673 cguest_time (time spent running a virtual CPU, see
674 below).
675 cstime %ld (17) Amount of time that this process's waited-for
677 children have been scheduled in kernel mode, mea‐
678 sured in clock ticks (divide by
679 sysconf(_SC_CLK_TCK)).
680 priority %ld
682 (18) (Explanation for Linux 2.6) For processes run‐
683 ning a real-time scheduling policy (policy below;
684 see sched_setscheduler(2)), this is the negated
685 scheduling priority, minus one; that is, a number in
686 the range -2 to -100, corresponding to real-time
687 priorities 1 to 99. For processes running under a
688 non-real-time scheduling policy, this is the raw
689 nice value (setpriority(2)) as represented in the
690 kernel. The kernel stores nice values as numbers in
691 the range 0 (high) to 39 (low), corresponding to the
692 user-visible nice range of -20 to 19.
693 Before Linux 2.6, this was a scaled value based on
695 the scheduler weighting given to this process.
696 nice %ld (19) The nice value (see setpriority(2)), a value in
698 the range 19 (low priority) to -20 (high priority).
699 num_threads %ld
701 (20) Number of threads in this process (since Linux
702 2.6). Before kernel 2.6, this field was hard coded
703 to 0 as a placeholder for an earlier removed field.
704 itrealvalue %ld
706 (21) The time in jiffies before the next SIGALRM is
707 sent to the process due to an interval timer. Since
708 kernel 2.6.17, this field is no longer maintained,
709 and is hard coded as 0.
710 starttime %llu (was %lu before Linux 2.6)
712 (22) The time the process started after system boot.
713 In kernels before Linux 2.6, this value was
714 expressed in jiffies. Since Linux 2.6, the value is
715 expressed in clock ticks (divide by
716 sysconf(_SC_CLK_TCK)).
717 vsize %lu (23) Virtual memory size in bytes.
719 rss %ld (24) Resident Set Size: number of pages the process
721 has in real memory. This is just the pages which
722 count toward text, data, or stack space. This does
723 not include pages which have not been demand-loaded
724 in, or which are swapped out.
725 rsslim %lu (25) Current soft limit in bytes on the rss of the
727 process; see the description of RLIMIT_RSS in getr‐
728 limit(2).
729 startcode %lu
731 (26) The address above which program text can run.
732 endcode %lu (27) The address below which program text can run.
734 startstack %lu
736 (28) The address of the start (i.e., bottom) of the
737 stack.
738 kstkesp %lu (29) The current value of ESP (stack pointer), as
740 found in the kernel stack page for the process.
741 kstkeip %lu (30) The current EIP (instruction pointer).
743 signal %lu (31) The bitmap of pending signals, displayed as a
745 decimal number. Obsolete, because it does not pro‐
746 vide information on real-time signals; use
747 /proc/[pid]/status instead.
748 blocked %lu (32) The bitmap of blocked signals, displayed as a
750 decimal number. Obsolete, because it does not pro‐
751 vide information on real-time signals; use
752 /proc/[pid]/status instead.
753 sigignore %lu
755 (33) The bitmap of ignored signals, displayed as a
756 decimal number. Obsolete, because it does not pro‐
757 vide information on real-time signals; use
758 /proc/[pid]/status instead.
759 sigcatch %lu
761 (34) The bitmap of caught signals, displayed as a
762 decimal number. Obsolete, because it does not pro‐
763 vide information on real-time signals; use
764 /proc/[pid]/status instead.
765 wchan %lu (35) This is the "channel" in which the process is
767 waiting. It is the address of a system call, and
768 can be looked up in a namelist if you need a textual
769 name. (If you have an up-to-date /etc/psdatabase,
770 then try ps -l to see the WCHAN field in action.)
771 nswap %lu (36) Number of pages swapped (not maintained).
773 cnswap %lu (37) Cumulative nswap for child processes (not main‐
775 tained).
776 exit_signal %d (since Linux 2.1.22)
778 (38) Signal to be sent to parent when we die.
779 processor %d (since Linux 2.2.8)
781 (39) CPU number last executed on.
782 rt_priority %u (since Linux 2.5.19; was %lu before Linux 2.6.22)
784 (40) Real-time scheduling priority, a number in the
785 range 1 to 99 for processes scheduled under a real-
786 time policy, or 0, for non-real-time processes (see
787 sched_setscheduler(2)).
788 policy %u (since Linux 2.5.19; was %lu before Linux 2.6.22)
790 (41) Scheduling policy (see sched_setscheduler(2)).
791 Decode using the SCHED_* constants in linux/sched.h.
792 delayacct_blkio_ticks %llu (since Linux 2.6.18)
794 (42) Aggregated block I/O delays, measured in clock
795 ticks (centiseconds).
796 guest_time %lu (since Linux 2.6.24)
798 (43) Guest time of the process (time spent running a
799 virtual CPU for a guest operating system), measured
800 in clock ticks (divide by sysconf(_SC_CLK_TCK)).
801 cguest_time %ld (since Linux 2.6.24)
803 (44) Guest time of the process's children, measured
804 in clock ticks (divide by sysconf(_SC_CLK_TCK)).
805 /proc/[pid]/statm
807 Provides information about memory usage, measured in pages. The
808 columns are:
809 size (1) total program size
811 (same as VmSize in /proc/[pid]/status)
812 resident (2) resident set size
813 (same as VmRSS in /proc/[pid]/status)
814 share (3) shared pages (i.e., backed by a file)
815 text (4) text (code)
816 lib (5) library (unused in Linux 2.6)
817 data (6) data + stack
818 dt (7) dirty pages (unused in Linux 2.6)
819 /proc/[pid]/status
821 Provides much of the information in /proc/[pid]/stat and
822 /proc/[pid]/statm in a format that's easier for humans to parse.
823 Here's an example:
824 $ cat /proc/$$/status
826 Name: bash
827 State: S (sleeping)
828 Tgid: 3515
829 Pid: 3515
830 PPid: 3452
831 TracerPid: 0
832 Uid: 1000 1000 1000 1000
833 Gid: 100 100 100 100
834 FDSize: 256
835 Groups: 16 33 100
836 VmPeak: 9136 kB
837 VmSize: 7896 kB
838 VmLck: 0 kB
839 VmHWM: 7572 kB
840 VmRSS: 6316 kB
841 VmData: 5224 kB
842 VmStk: 88 kB
843 VmExe: 572 kB
844 VmLib: 1708 kB
845 VmPTE: 20 kB
846 Threads: 1
847 SigQ: 0/3067
848 SigPnd: 0000000000000000
849 ShdPnd: 0000000000000000
850 SigBlk: 0000000000010000
851 SigIgn: 0000000000384004
852 SigCgt: 000000004b813efb
853 CapInh: 0000000000000000
854 CapPrm: 0000000000000000
855 CapEff: 0000000000000000
856 CapBnd: ffffffffffffffff
857 Cpus_allowed: 00000001
858 Cpus_allowed_list: 0
859 Mems_allowed: 1
860 Mems_allowed_list: 0
861 voluntary_ctxt_switches: 150
862 nonvoluntary_ctxt_switches: 545
863 The fields are as follows:
865 * Name: Command run by this process.
867 * State: Current state of the process. One of "R (running)", "S
869 (sleeping)", "D (disk sleep)", "T (stopped)", "T (tracing
870 stop)", "Z (zombie)", or "X (dead)".
871 * Tgid: Thread group ID (i.e., Process ID).
873 * Pid: Thread ID (see gettid(2)).
875 * PPid: PID of parent process.
877 * TracerPid: PID of process tracing this process (0 if not being
879 traced).
880 * Uid, Gid: Real, effective, saved set, and file system UIDs
882 (GIDs).
883 * FDSize: Number of file descriptor slots currently allocated.
885 * Groups: Supplementary group list.
887 * VmPeak: Peak virtual memory size.
889 * VmSize: Virtual memory size.
891 * VmLck: Locked memory size (see mlock(3)).
893 * VmHWM: Peak resident set size ("high water mark").
895 * VmRSS: Resident set size.
897 * VmData, VmStk, VmExe: Size of data, stack, and text segments.
899 * VmLib: Shared library code size.
901 * VmPTE: Page table entries size (since Linux 2.6.10).
903 * Threads: Number of threads in process containing this thread.
905 * SigQ: This field contains two slash-separated numbers that
907 relate to queued signals for the real user ID of this process.
908 The first of these is the number of currently queued signals
909 for this real user ID, and the second is the resource limit on
910 the number of queued signals for this process (see the
911 description of RLIMIT_SIGPENDING in getrlimit(2)).
912 * SigPnd, ShdPnd: Number of signals pending for thread and for
914 process as a whole (see pthreads(7) and signal(7)).
915 * SigBlk, SigIgn, SigCgt: Masks indicating signals being
917 blocked, ignored, and caught (see signal(7)).
918 * CapInh, CapPrm, CapEff: Masks of capabilities enabled in
920 inheritable, permitted, and effective sets (see capabili‐
921 ties(7)).
922 * CapBnd: Capability Bounding set (since kernel 2.6.26, see
924 capabilities(7)).
925 * Cpus_allowed: Mask of CPUs on which this process may run
927 (since Linux 2.6.24, see cpuset(7)).
928 * Cpus_allowed_list: Same as previous, but in "list format"
930 (since Linux 2.6.26, see cpuset(7)).
931 * Mems_allowed: Mask of memory nodes allowed to this process
933 (since Linux 2.6.24, see cpuset(7)).
934 * Mems_allowed_list: Same as previous, but in "list format"
936 (since Linux 2.6.26, see cpuset(7)).
937 * voluntary_context_switches, nonvoluntary_context_switches:
939 Number of voluntary and involuntary context switches (since
940 Linux 2.6.23).
941 /proc/[pid]/task (since Linux 2.6.0-test6)
943 This is a directory that contains one subdirectory for each
944 thread in the process. The name of each subdirectory is the
945 numerical thread ID ([tid]) of the thread (see gettid(2)).
946 Within each of these subdirectories, there is a set of files
947 with the same names and contents as under the /proc/[pid] direc‐
948 tories. For attributes that are shared by all threads, the con‐
949 tents for each of the files under the task/[tid] subdirectories
950 will be the same as in the corresponding file in the parent
951 /proc/[pid] directory (e.g., in a multithreaded process, all of
952 the task/[tid]/cwd files will have the same value as the
953 /proc/[pid]/cwd file in the parent directory, since all of the
954 threads in a process share a working directory). For attributes
955 that are distinct for each thread, the corresponding files under
956 task/[tid] may have different values (e.g., various fields in
957 each of the task/[tid]/status files may be different for each
958 thread).
959 In a multithreaded process, the contents of the /proc/[pid]/task
961 directory are not available if the main thread has already ter‐
962 minated (typically by calling pthread_exit(3)).
963 /proc/apm
965 Advanced power management version and battery information when
966 CONFIG_APM is defined at kernel compilation time.
967 /proc/bus
969 Contains subdirectories for installed busses.
970 /proc/bus/pccard
972 Subdirectory for PCMCIA devices when CONFIG_PCMCIA is set at
973 kernel compilation time.
974 /proc/bus/pccard/drivers
976 /proc/bus/pci
978 Contains various bus subdirectories and pseudo-files containing
979 information about PCI busses, installed devices, and device
980 drivers. Some of these files are not ASCII.
981 /proc/bus/pci/devices
983 Information about PCI devices. They may be accessed through
984 lspci(8) and setpci(8).
985 /proc/cmdline
987 Arguments passed to the Linux kernel at boot time. Often done
988 via a boot manager such as lilo(8) or grub(8).
989 /proc/config.gz (since Linux 2.6)
991 This file exposes the configuration options that were used to
992 build the currently running kernel, in the same format as they
993 would be shown in the .config file that resulted when configur‐
994 ing the kernel (using make xconfig, make config, or similar).
995 The file contents are compressed; view or search them using
996 zcat(1), zgrep(1), etc. As long as no changes have been made to
997 the following file, the contents of /proc/config.gz are the same
998 as those provided by :
999 cat /lib/modules/$(uname -r)/build/.config
1001 /proc/config.gz is provided only if the kernel is configured
1003 with CONFIG_IKCONFIG_PROC.
1004 /proc/cpuinfo
1006 This is a collection of CPU and system architecture dependent
1007 items, for each supported architecture a different list. Two
1008 common entries are processor which gives CPU number and
1009 bogomips; a system constant that is calculated during kernel
1010 initialization. SMP machines have information for each CPU.
1011 The lscpu(1) command gathers its information from this file.
1012 /proc/devices
1014 Text listing of major numbers and device groups. This can be
1015 used by MAKEDEV scripts for consistency with the kernel.
1016 /proc/diskstats (since Linux 2.5.69)
1018 This file contains disk I/O statistics for each disk device.
1019 See the Linux kernel source file Documentation/iostats.txt for
1020 further information.
1021 /proc/dma
1023 This is a list of the registered ISA DMA (direct memory access)
1024 channels in use.
1025 /proc/driver
1027 Empty subdirectory.
1028 /proc/execdomains
1030 List of the execution domains (ABI personalities).
1031 /proc/fb
1033 Frame buffer information when CONFIG_FB is defined during kernel
1034 compilation.
1035 /proc/filesystems
1037 A text listing of the file systems which are supported by the
1038 kernel, namely file systems which were compiled into the kernel
1039 or whose kernel modules are currently loaded. (See also
1040 filesystems(5).) If a file system is marked with "nodev", this
1041 means that it does not require a block device to be mounted
1042 (e.g., virtual file system, network file system).
1043 Incidentally, this file may be used by mount(8) when no file
1045 system is specified and it didn't manage to determine the file
1046 system type. Then file systems contained in this file are tried
1047 (excepted those that are marked with "nodev").
1048 /proc/fs
1050 Empty subdirectory.
1051 /proc/ide
1053 This directory exists on systems with the IDE bus. There are
1054 directories for each IDE channel and attached device. Files
1055 include:
1056 cache buffer size in KB
1058 capacity number of sectors
1059 driver driver version
1060 geometry physical and logical geometry
1061 identify in hexadecimal
1062 media media type
1063 model manufacturer's model number
1064 settings drive settings
1065 smart_thresholds in hexadecimal
1066 smart_values in hexadecimal
1067 The hdparm(8) utility provides access to this information in a
1069 friendly format.
1070 /proc/interrupts
1072 This is used to record the number of interrupts per CPU per IO
1073 device. Since Linux 2.6.24, for the i386 and x86_64 architec‐
1074 tures, at least, this also includes interrupts internal to the
1075 system (that is, not associated with a device as such), such as
1076 NMI (nonmaskable interrupt), LOC (local timer interrupt), and
1077 for SMP systems, TLB (TLB flush interrupt), RES (rescheduling
1078 interrupt), CAL (remote function call interrupt), and possibly
1079 others. Very easy to read formatting, done in ASCII.
1080 /proc/iomem
1082 I/O memory map in Linux 2.4.
1083 /proc/ioports
1085 This is a list of currently registered Input-Output port regions
1086 that are in use.
1087 /proc/kallsyms (since Linux 2.5.71)
1089 This holds the kernel exported symbol definitions used by the
1090 modules(X) tools to dynamically link and bind loadable modules.
1091 In Linux 2.5.47 and earlier, a similar file with slightly dif‐
1092 ferent syntax was named ksyms.
1093 /proc/kcore
1095 This file represents the physical memory of the system and is
1096 stored in the ELF core file format. With this pseudo-file, and
1097 an unstripped kernel (/usr/src/linux/vmlinux) binary, GDB can be
1098 used to examine the current state of any kernel data structures.
1099 The total length of the file is the size of physical memory
1101 (RAM) plus 4KB.
1102 /proc/kmsg
1104 This file can be used instead of the syslog(2) system call to
1105 read kernel messages. A process must have superuser privileges
1106 to read this file, and only one process should read this file.
1107 This file should not be read if a syslog process is running
1108 which uses the syslog(2) system call facility to log kernel mes‐
1109 sages.
1110 Information in this file is retrieved with the dmesg(1) program.
1112 /proc/ksyms (Linux 1.1.23-2.5.47)
1114 See /proc/kallsyms.
1115 /proc/loadavg
1117 The first three fields in this file are load average figures
1118 giving the number of jobs in the run queue (state R) or waiting
1119 for disk I/O (state D) averaged over 1, 5, and 15 minutes. They
1120 are the same as the load average numbers given by uptime(1) and
1121 other programs. The fourth field consists of two numbers sepa‐
1122 rated by a slash (/). The first of these is the number of cur‐
1123 rently runnable kernel scheduling entities (processes, threads).
1124 The value after the slash is the number of kernel scheduling
1125 entities that currently exist on the system. The fifth field is
1126 the PID of the process that was most recently created on the
1127 system.
1128 /proc/locks
1130 This file shows current file locks (flock(2) and fcntl(2)) and
1131 leases (fcntl(2)).
1132 /proc/malloc (only up to and including Linux 2.2)
1134 This file is present only if CONFIG_DEBUG_MALLOC was defined
1135 during compilation.
1136 /proc/meminfo
1138 This file reports statistics about memory usage on the system.
1139 It is used by free(1) to report the amount of free and used mem‐
1140 ory (both physical and swap) on the system as well as the shared
1141 memory and buffers used by the kernel. Each line of the file
1142 consists of a parameter name, followed by a colon, the value of
1143 the parameter, and an option unit of measurement (e.g., "kB").
1144 The list below describes the parameter names and the format
1145 specifier required to read the field value. Except as noted
1146 below, all of the fields have been present since at least Linux
1147 2.6.0. Some fileds are displayed only if the kernel was config‐
1148 ured with various options; those dependencies are noted in the
1149 list.
1150 MemTotal %lu
1152 Total usable RAM (i.e. physical RAM minus a few reserved
1153 bits and the kernel binary code).
1154 MemFree %lu
1156 The sum of LowFree+HighFree.
1157 Buffers %lu
1159 Relatively temporary storage for raw disk blocks that
1160 shouldn't get tremendously large (20MB or so).
1161 Cached %lu
1163 In-memory cache for files read from the disk (the page
1164 cache). Doesn't include SwapCached.
1165 SwapCached %lu
1167 Memory that once was swapped out, is swapped back in but
1168 still also is in the swap file. (If memory pressure is
1169 high, these pages don't need to be swapped out again
1170 because they are already in the swap file. This saves
1171 I/O.)
1172 Active %lu
1174 Memory that has been used more recently and usually not
1175 reclaimed unless absolutely necessary.
1176 Inactive %lu
1178 Memory which has been less recently used. It is more
1179 eligible to be reclaimed for other purposes.
1180 Active(anon) %lu (since Linux 2.6.28)
1182 [To be documented.]
1183 Inactive(anon) %lu (since Linux 2.6.28)
1185 [To be documented.]
1186 Active(file) %lu (since Linux 2.6.28)
1188 [To be documented.]
1189 Inactive(file) %lu (since Linux 2.6.28)
1191 [To be documented.]
1192 Unevictable %lu (since Linux 2.6.28)
1194 (From Linux 2.6.28 to 2.6.30, CONFIG_UNEVICTABLE_LRU was
1195 required.) [To be documented.]
1196 Mlocked %lu (since Linux 2.6.28)
1198 (From Linux 2.6.28 to 2.6.30, CONFIG_UNEVICTABLE_LRU was
1199 required.) [To be documented.]
1200 HighTotal %lu
1202 (Starting with Linux 2.6.19, CONFIG_HIGHMEM is required.)
1203 Total amount of highmem. Highmem is all memory above
1204 ~860MB of physical memory. Highmem areas are for use by
1205 user-space programs, or for the page cache. The kernel
1206 must use tricks to access this memory, making it slower
1207 to access than lowmem.
1208 HighFree %lu
1210 (Starting with Linux 2.6.19, CONFIG_HIGHMEM is required.)
1211 Amount of free highmem.
1212 LowTotal %lu
1214 (Starting with Linux 2.6.19, CONFIG_HIGHMEM is required.)
1215 Total amount of lowmem. Lowmem is memory which can be
1216 used for everything that highmem can be used for, but it
1217 is also available for the kernel's use for its own data
1218 structures. Among many other things, it is where every‐
1219 thing from Slab is allocated. Bad things happen when
1220 you're out of lowmem.
1221 LowFree %lu
1223 (Starting with Linux 2.6.19, CONFIG_HIGHMEM is required.)
1224 Amount of free lowmem.
1225 MmapCopy %lu (since Linux 2.6.29)
1227 (CONFIG_MMU is required.) [To be documented.]
1228 SwapTotal %lu
1230 Total amount of swap space available.
1231 SwapFree %lu
1233 Amount of swap space that is currently unused.
1234 Dirty %lu
1236 Memory which is waiting to get written back to the disk.
1237 Writeback %lu
1239 Memory which is actively being written back to the disk.
1240 AnonPages %lu (since Linux 2.6.18)
1242 Non-file backed pages mapped into user-space page tables.
1243 Mapped %lu
1245 Files which have been mmaped, such as libraries.
1246 Shmem %lu (since Linux 2.6.32)
1248 [To be documented.]
1249 Slab %lu
1251 In-kernel data structures cache.
1252 SReclaimable %lu (since Linux 2.6.19)
1254 Part of Slab, that might be reclaimed, such as caches.
1255 SUnreclaim %lu (since Linux 2.6.19)
1257 Part of Slab, that cannot be reclaimed on memory pres‐
1258 sure.
1259 KernelStack %lu (since Linux 2.6.32)
1261 Amount of memory allocated to kernel stacks.
1262 PageTables %lu (since Linux 2.6.18)
1264 Amount of memory dedicated to the lowest level of page
1265 tables.
1266 Quicklists %lu (since Linux 2.6.27)
1268 (CONFIG_QUICKLIST is required.) [To be documented.]
1269 NFS_Unstable %lu (since Linux 2.6.18)
1271 NFS pages sent to the server, but not yet committed to
1272 stable storage.
1273 Bounce %lu (since Linux 2.6.18)
1275 Memory used for block device "bounce buffers".
1276 WritebackTmp %lu (since Linux 2.6.26)
1278 Memory used by FUSE for temporary writeback buffers.
1279 CommitLimit %lu (since Linux 2.6.10)
1281 Based on the overcommit ratio ('vm.overcommit_ratio'),
1282 this is the total amount of memory currently available
1283 to be allocated on the system. This limit is adhered to
1284 only if strict overcommit accounting is enabled (mode 2
1285 in /proc/sys/vm/overcommit_ratio). The CommitLimit is
1286 calculated using the following formula:
1287 CommitLimit = (overcommit_ratio * Physical RAM) +
1289 Swap
1290 For example, on a system with 1GB of physical RAM and 7GB
1292 of swap with a overcommit_ratio of 30, this formula
1293 yields a CommitLimit of 7.3GB. For more details, see the
1294 memory overcommit documentation in the kernel source file
1295 Documentation/vm/overcommit-accounting.
1296 Committed_AS %lu
1298 The amount of memory presently allocated on the system.
1299 The committed memory is a sum of all of the memory which
1300 has been allocated by processes, even if it has not been
1301 "used" by them as of yet. A process which allocates 1GB
1302 of memory (using malloc(3) or similar), but touches only
1303 300MB of that memory will show up as using only 300MB of
1304 memory even if it has the address space allocated for the
1305 entire 1GB. This 1GB is memory which has been "commit‐
1306 ted" to by the VM and can be used at any time by the
1307 allocating application. With strict overcommit enabled
1308 on the system (mode 2 /proc/sys/vm/overcommit_memory),
1309 allocations which would exceed the CommitLimit (detailed
1310 above) will not be permitted. This is useful if one
1311 needs to guarantee that processes will not fail due to
1312 lack of memory once that memory has been successfully
1313 allocated.
1314 VmallocTotal %lu
1316 Total size of vmalloc memory area.
1317 VmallocUsed %lu
1319 Amount of vmalloc area which is used.
1320 VmallocChunk %lu
1322 Largest contiguous block of vmalloc area which is free.
1323 HardwareCorrupted %lu (since Linux 2.6.32)
1325 (CONFIG_MEMORY_FAILURE is required.) [To be documented.]
1326 AnonHugePages %lu (since Linux 2.6.38)
1328 (CONFIG_TRANSPARENT_HUGEPAGE is required.) Non-file
1329 backed huge pages mapped into user-space page tables.
1330 HugePages_Total %lu
1332 (CONFIG_HUGETLB_PAGE is required.) The size of the pool
1333 of huge pages.
1334 HugePages_Free %lu
1336 (CONFIG_HUGETLB_PAGE is required.) The number of huge
1337 pages in the pool that are not yet allocated.
1338 HugePages_Rsvd %lu (since Linux 2.6.17)
1340 (CONFIG_HUGETLB_PAGE is required.) This is the number of
1341 huge pages for which a commitment to allocate from the
1342 pool has been made, but no allocation has yet been made.
1343 These reserved huge pages guarantee that an application
1344 will be able to allocate a huge page from the pool of
1345 huge pages at fault time.
1346 HugePages_Surp %lu (since Linux 2.6.24)
1348 (CONFIG_HUGETLB_PAGE is required.) This is the number of
1349 huge pages in the pool above the value in
1350 /proc/sys/vm/nr_hugepages. The maximum number of surplus
1351 huge pages is controlled by /proc/sys/vm/nr_overcom‐
1352 mit_hugepages.
1353 Hugepagesize %lu
1355 (CONFIG_HUGETLB_PAGE is required.) The size of huge
1356 pages.
1357 /proc/modules
1359 A text list of the modules that have been loaded by the system.
1360 See also lsmod(8).
1361 /proc/mounts
1363 Before kernel 2.4.19, this file was a list of all the file sys‐
1364 tems currently mounted on the system. With the introduction of
1365 per-process mount namespaces in Linux 2.4.19, this file became a
1366 link to /proc/self/mounts, which lists the mount points of the
1367 process's own mount namespace. The format of this file is docu‐
1368 mented in fstab(5).
1369 /proc/mtrr
1371 Memory Type Range Registers. See the Linux kernel source file
1372 Documentation/mtrr.txt for details.
1373 /proc/net
1375 various net pseudo-files, all of which give the status of some
1376 part of the networking layer. These files contain ASCII struc‐
1377 tures and are, therefore, readable with cat(1). However, the
1378 standard netstat(8) suite provides much cleaner access to these
1379 files.
1380 /proc/net/arp
1382 This holds an ASCII readable dump of the kernel ARP table used
1383 for address resolutions. It will show both dynamically learned
1384 and preprogrammed ARP entries. The format is:
1385 IP address HW type Flags HW address Mask Device
1387 192.168.0.50 0x1 0x2 00:50:BF:25:68:F3 * eth0
1388 192.168.0.250 0x1 0xc 00:00:00:00:00:00 * eth0
1389 Here "IP address" is the IPv4 address of the machine and the "HW
1391 type" is the hardware type of the address from RFC 826. The
1392 flags are the internal flags of the ARP structure (as defined in
1393 /usr/include/linux/if_arp.h) and the "HW address" is the data
1394 link layer mapping for that IP address if it is known.
1395 /proc/net/dev
1397 The dev pseudo-file contains network device status information.
1398 This gives the number of received and sent packets, the number
1399 of errors and collisions and other basic statistics. These are
1400 used by the ifconfig(8) program to report device status. The
1401 format is:
1402 Inter-| Receive | Transmit
1404 face |bytes packets errs drop fifo frame compressed multicast|bytes packets errs drop fifo colls carrier compressed
1405 lo: 2776770 11307 0 0 0 0 0 0 2776770 11307 0 0 0 0 0 0
1406 eth0: 1215645 2751 0 0 0 0 0 0 1782404 4324 0 0 0 427 0 0
1407 ppp0: 1622270 5552 1 0 0 0 0 0 354130 5669 0 0 0 0 0 0
1408 tap0: 7714 81 0 0 0 0 0 0 7714 81 0 0 0 0 0 0
1409 /proc/net/dev_mcast
1411 Defined in /usr/src/linux/net/core/dev_mcast.c:
1412 indx interface_name dmi_u dmi_g dmi_address
1413 2 eth0 1 0 01005e000001
1414 3 eth1 1 0 01005e000001
1415 4 eth2 1 0 01005e000001
1416 /proc/net/igmp
1418 Internet Group Management Protocol. Defined in
1419 /usr/src/linux/net/core/igmp.c.
1420 /proc/net/rarp
1422 This file uses the same format as the arp file and contains the
1423 current reverse mapping database used to provide rarp(8) reverse
1424 address lookup services. If RARP is not configured into the
1425 kernel, this file will not be present.
1426 /proc/net/raw
1428 Holds a dump of the RAW socket table. Much of the information
1429 is not of use apart from debugging. The "sl" value is the ker‐
1430 nel hash slot for the socket, the "local_address" is the local
1431 address and protocol number pair. "St" is the internal status
1432 of the socket. The "tx_queue" and "rx_queue" are the outgoing
1433 and incoming data queue in terms of kernel memory usage. The
1434 "tr", "tm->when", and "rexmits" fields are not used by RAW. The
1435 "uid" field holds the effective UID of the creator of the
1436 socket.
1437 /proc/net/snmp
1439 This file holds the ASCII data needed for the IP, ICMP, TCP, and
1440 UDP management information bases for an SNMP agent.
1441 /proc/net/tcp
1443 Holds a dump of the TCP socket table. Much of the information
1444 is not of use apart from debugging. The "sl" value is the ker‐
1445 nel hash slot for the socket, the "local_address" is the local
1446 address and port number pair. The "rem_address" is the remote
1447 address and port number pair (if connected). "St" is the inter‐
1448 nal status of the socket. The "tx_queue" and "rx_queue" are the
1449 outgoing and incoming data queue in terms of kernel memory
1450 usage. The "tr", "tm->when", and "rexmits" fields hold internal
1451 information of the kernel socket state and are only useful for
1452 debugging. The "uid" field holds the effective UID of the cre‐
1453 ator of the socket.
1454 /proc/net/udp
1456 Holds a dump of the UDP socket table. Much of the information
1457 is not of use apart from debugging. The "sl" value is the ker‐
1458 nel hash slot for the socket, the "local_address" is the local
1459 address and port number pair. The "rem_address" is the remote
1460 address and port number pair (if connected). "St" is the inter‐
1461 nal status of the socket. The "tx_queue" and "rx_queue" are the
1462 outgoing and incoming data queue in terms of kernel memory
1463 usage. The "tr", "tm->when", and "rexmits" fields are not used
1464 by UDP. The "uid" field holds the effective UID of the creator
1465 of the socket. The format is:
1466 sl local_address rem_address st tx_queue rx_queue tr rexmits tm->when uid
1468 1: 01642C89:0201 0C642C89:03FF 01 00000000:00000001 01:000071BA 00000000 0
1469 1: 00000000:0801 00000000:0000 0A 00000000:00000000 00:00000000 6F000100 0
1470 1: 00000000:0201 00000000:0000 0A 00000000:00000000 00:00000000 00000000 0
1471 /proc/net/unix
1473 Lists the UNIX domain sockets present within the system and
1474 their status. The format is:
1475 Num RefCount Protocol Flags Type St Path
1476 0: 00000002 00000000 00000000 0001 03
1477 1: 00000001 00000000 00010000 0001 01 /dev/printer
1478 Here "Num" is the kernel table slot number, "RefCount" is the
1480 number of users of the socket, "Protocol" is currently always 0,
1481 "Flags" represent the internal kernel flags holding the status
1482 of the socket. Currently, type is always "1" (UNIX domain data‐
1483 gram sockets are not yet supported in the kernel). "St" is the
1484 internal state of the socket and Path is the bound path (if any)
1485 of the socket.
1486 /proc/partitions
1488 Contains major and minor numbers of each partition as well as
1489 number of blocks and partition name.
1490 /proc/pci
1492 This is a listing of all PCI devices found during kernel ini‐
1493 tialization and their configuration.
1494 This file has been deprecated in favor of a new /proc interface
1496 for PCI (/proc/bus/pci). It became optional in Linux 2.2
1497 (available with CONFIG_PCI_OLD_PROC set at kernel compilation).
1498 It became once more nonoptionally enabled in Linux 2.4. Next,
1499 it was deprecated in Linux 2.6 (still available with CON‐
1500 FIG_PCI_LEGACY_PROC set), and finally removed altogether since
1501 Linux 2.6.17.
1502 /proc/profile (since Linux 2.4)
1504 This file is present only if the kernel was booted with the pro‐
1505 file=1 command-line option. It exposes kernel profiling infor‐
1506 mation in a binary format for use by readprofile(1). Writing
1507 (e.g., an empty string) to this file resets the profiling coun‐
1508 ters; on some architectures, writing a binary integer "profiling
1509 multiplier" of size sizeof(int) sets the profiling interrupt
1510 frequency.
1511 /proc/scsi
1513 A directory with the scsi mid-level pseudo-file and various SCSI
1514 low-level driver directories, which contain a file for each SCSI
1515 host in this system, all of which give the status of some part
1516 of the SCSI IO subsystem. These files contain ASCII structures
1517 and are, therefore, readable with cat(1).
1518 You can also write to some of the files to reconfigure the sub‐
1520 system or switch certain features on or off.
1521 /proc/scsi/scsi
1523 This is a listing of all SCSI devices known to the kernel. The
1524 listing is similar to the one seen during bootup. scsi cur‐
1525 rently supports only the add-single-device command which allows
1526 root to add a hotplugged device to the list of known devices.
1527 The command
1529 echo 'scsi add-single-device 1 0 5 0' > /proc/scsi/scsi
1531 will cause host scsi1 to scan on SCSI channel 0 for a device on
1533 ID 5 LUN 0. If there is already a device known on this address
1534 or the address is invalid, an error will be returned.
1535 /proc/scsi/[drivername]
1537 [drivername] can currently be NCR53c7xx, aha152x, aha1542,
1538 aha1740, aic7xxx, buslogic, eata_dma, eata_pio, fdomain, in2000,
1539 pas16, qlogic, scsi_debug, seagate, t128, u15-24f, ultrastore,
1540 or wd7000. These directories show up for all drivers that reg‐
1541 istered at least one SCSI HBA. Every directory contains one
1542 file per registered host. Every host-file is named after the
1543 number the host was assigned during initialization.
1544 Reading these files will usually show driver and host configura‐
1546 tion, statistics, etc.
1547 Writing to these files allows different things on different
1549 hosts. For example, with the latency and nolatency commands,
1550 root can switch on and off command latency measurement code in
1551 the eata_dma driver. With the lockup and unlock commands, root
1552 can control bus lockups simulated by the scsi_debug driver.
1553 /proc/self
1555 This directory refers to the process accessing the /proc file
1556 system, and is identical to the /proc directory named by the
1557 process ID of the same process.
1558 /proc/slabinfo
1560 Information about kernel caches. Since Linux 2.6.16 this file
1561 is present only if the CONFIG_SLAB kernel configuration option
1562 is enabled. The columns in /proc/slabinfo are:
1563 cache-name
1565 num-active-objs
1566 total-objs
1567 object-size
1568 num-active-slabs
1569 total-slabs
1570 num-pages-per-slab
1571 See slabinfo(5) for details.
1573 /proc/stat
1575 kernel/system statistics. Varies with architecture. Common
1576 entries include:
1577 cpu 3357 0 4313 1362393
1579 The amount of time, measured in units of USER_HZ
1580 (1/100ths of a second on most architectures, use
1581 sysconf(_SC_CLK_TCK) to obtain the right value), that the
1582 system spent in various states:
1583 user (1) Time spent in user mode.
1585 nice (2) Time spent in user mode with low priority
1587 (nice).
1588 system (3) Time spent in system mode.
1590 idle (4) Time spent in the idle task. This value
1592 should be USER_HZ times the second entry in the
1593 /proc/uptime pseudo-file.
1594 iowait (since Linux 2.5.41)
1596 (5) Time waiting for I/O to complete.
1597 irq (since Linux 2.6.0-test4)
1599 (6) Time servicing interrupts.
1600 softirq (since Linux 2.6.0-test4)
1602 (7) Time servicing softirqs.
1603 steal (since Linux 2.6.11)
1605 (8) Stolen time, which is the time spent in other
1606 operating systems when running in a virtualized
1607 environment
1608 guest (since Linux 2.6.24)
1610 (9) Time spent running a virtual CPU for guest
1611 operating systems under the control of the Linux
1612 kernel.
1613 guest_nice (since Linux 2.6.33)
1615 (10) Time spent running a niced guest (virtual CPU
1616 for guest operating systems under the control of
1617 the Linux kernel).
1618 page 5741 1808
1620 The number of pages the system paged in and the number
1621 that were paged out (from disk).
1622 swap 1 0
1624 The number of swap pages that have been brought in and
1625 out.
1626 intr 1462898
1628 This line shows counts of interrupts serviced since boot
1629 time, for each of the possible system interrupts. The
1630 first column is the total of all interrupts serviced;
1631 each subsequent column is the total for a particular
1632 interrupt.
1633 disk_io: (2,0):(31,30,5764,1,2) (3,0):...
1635 (major,disk_idx):(noinfo, read_io_ops, blks_read,
1636 write_io_ops, blks_written)
1637 (Linux 2.4 only)
1638 ctxt 115315
1640 The number of context switches that the system underwent.
1641 btime 769041601
1643 boot time, in seconds since the Epoch, 1970-01-01
1644 00:00:00 +0000 (UTC).
1645 processes 86031
1647 Number of forks since boot.
1648 procs_running 6
1650 Number of processes in runnable state. (Linux 2.5.45
1651 onward.)
1652 procs_blocked 2
1654 Number of processes blocked waiting for I/O to complete.
1655 (Linux 2.5.45 onward.)
1656 /proc/swaps
1658 Swap areas in use. See also swapon(8).
1659 /proc/sys
1661 This directory (present since 1.3.57) contains a number of files
1662 and subdirectories corresponding to kernel variables. These
1663 variables can be read and sometimes modified using the /proc
1664 file system, and the (deprecated) sysctl(2) system call.
1665 /proc/sys/abi (since Linux 2.4.10)
1667 This directory may contain files with application binary infor‐
1668 mation. See the Linux kernel source file Documenta‐
1669 tion/sysctl/abi.txt for more information.
1670 /proc/sys/debug
1672 This directory may be empty.
1673 /proc/sys/dev
1675 This directory contains device-specific information (e.g.,
1676 dev/cdrom/info). On some systems, it may be empty.
1677 /proc/sys/fs
1679 This directory contains the files and subdirectories for kernel
1680 variables related to file systems.
1681 /proc/sys/fs/binfmt_misc
1683 Documentation for files in this directory can be found in the
1684 Linux kernel sources in Documentation/binfmt_misc.txt.
1685 /proc/sys/fs/dentry-state (since Linux 2.2)
1687 This file contains information about the status of the directory
1688 cache (dcache). The file contains six numbers, nr_dentry,
1689 nr_unused, age_limit (age in seconds), want_pages (pages
1690 requested by system) and two dummy values.
1691 * nr_dentry is the number of allocated dentries (dcache
1693 entries). This field is unused in Linux 2.2.
1694 * nr_unused is the number of unused dentries.
1696 * age_limit is the age in seconds after which dcache entries can
1698 be reclaimed when memory is short.
1699 * want_pages is nonzero when the kernel has called
1701 shrink_dcache_pages() and the dcache isn't pruned yet.
1702 /proc/sys/fs/dir-notify-enable
1704 This file can be used to disable or enable the dnotify interface
1705 described in fcntl(2) on a system-wide basis. A value of 0 in
1706 this file disables the interface, and a value of 1 enables it.
1707 /proc/sys/fs/dquot-max
1709 This file shows the maximum number of cached disk quota entries.
1710 On some (2.4) systems, it is not present. If the number of free
1711 cached disk quota entries is very low and you have some awesome
1712 number of simultaneous system users, you might want to raise the
1713 limit.
1714 /proc/sys/fs/dquot-nr
1716 This file shows the number of allocated disk quota entries and
1717 the number of free disk quota entries.
1718 /proc/sys/fs/epoll (since Linux 2.6.28)
1720 This directory contains the file max_user_watches, which can be
1721 used to limit the amount of kernel memory consumed by the epoll
1722 interface. For further details, see epoll(7).
1723 /proc/sys/fs/file-max
1725 This file defines a system-wide limit on the number of open
1726 files for all processes. (See also setrlimit(2), which can be
1727 used by a process to set the per-process limit, RLIMIT_NOFILE,
1728 on the number of files it may open.) If you get lots of error
1729 messages in the kernel log about running out of file handles
1730 (look for "VFS: file-max limit <number> reached"), try increas‐
1731 ing this value:
1732 echo 100000 > /proc/sys/fs/file-max
1734 The kernel constant NR_OPEN imposes an upper limit on the value
1736 that may be placed in file-max.
1737 If you increase /proc/sys/fs/file-max, be sure to increase
1739 /proc/sys/fs/inode-max to 3-4 times the new value of
1740 /proc/sys/fs/file-max, or you will run out of inodes.
1741 Privileged processes (CAP_SYS_ADMIN) can override the file-max
1743 limit.
1744 /proc/sys/fs/file-nr
1746 This (read-only) file contains three numbers: the number of
1747 allocated file handles (i.e., the number of files presently
1748 opened); the number of free file handles; and the maximum number
1749 of file handles (i.e., the same value as /proc/sys/fs/file-max).
1750 If the number of allocated file handles is close to the maximum,
1751 you should consider increasing the maximum. Before Linux 2.6,
1752 the kernel allocated file handles dynamically, but it didn't
1753 free them again. Instead the free file handles were kept in a
1754 list for reallocation; the "free file handles" value indicates
1755 the size of that list. A large number of free file handles
1756 indicates that there was a past peak in the usage of open file
1757 handles. Since Linux 2.6, the kernel does deallocate freed file
1758 handles, and the "free file handles" value is always zero.
1759 /proc/sys/fs/inode-max
1761 This file contains the maximum number of in-memory inodes. On
1762 some (2.4) systems, it may not be present. This value should be
1763 3-4 times larger than the value in file-max, since stdin, stdout
1764 and network sockets also need an inode to handle them. When you
1765 regularly run out of inodes, you need to increase this value.
1766 /proc/sys/fs/inode-nr
1768 This file contains the first two values from inode-state.
1769 /proc/sys/fs/inode-state
1771 This file contains seven numbers: nr_inodes, nr_free_inodes,
1772 preshrink, and four dummy values. nr_inodes is the number of
1773 inodes the system has allocated. This can be slightly more than
1774 inode-max because Linux allocates them one page full at a time.
1775 nr_free_inodes represents the number of free inodes. preshrink
1776 is nonzero when the nr_inodes > inode-max and the system needs
1777 to prune the inode list instead of allocating more.
1778 /proc/sys/fs/inotify (since Linux 2.6.13)
1780 This directory contains files max_queued_events,
1781 max_user_instances, and max_user_watches, that can be used to
1782 limit the amount of kernel memory consumed by the inotify inter‐
1783 face. For further details, see inotify(7).
1784 /proc/sys/fs/lease-break-time
1786 This file specifies the grace period that the kernel grants to a
1787 process holding a file lease (fcntl(2)) after it has sent a sig‐
1788 nal to that process notifying it that another process is waiting
1789 to open the file. If the lease holder does not remove or down‐
1790 grade the lease within this grace period, the kernel forcibly
1791 breaks the lease.
1792 /proc/sys/fs/leases-enable
1794 This file can be used to enable or disable file leases
1795 (fcntl(2)) on a system-wide basis. If this file contains the
1796 value 0, leases are disabled. A nonzero value enables leases.
1797 /proc/sys/fs/mqueue (since Linux 2.6.6)
1799 This directory contains files msg_max, msgsize_max, and
1800 queues_max, controlling the resources used by POSIX message
1801 queues. See mq_overview(7) for details.
1802 /proc/sys/fs/overflowgid and /proc/sys/fs/overflowuid
1804 These files allow you to change the value of the fixed UID and
1805 GID. The default is 65534. Some file systems support only
1806 16-bit UIDs and GIDs, although in Linux UIDs and GIDs are 32
1807 bits. When one of these file systems is mounted with writes
1808 enabled, any UID or GID that would exceed 65535 is translated to
1809 the overflow value before being written to disk.
1810 /proc/sys/fs/pipe-max-size (since Linux 2.6.35)
1812 The value in this file defines an upper limit for raising the
1813 capacity of a pipe using the fcntl(2) F_SETPIPE_SZ operation.
1814 This limit applies only to unprivileged processes. The default
1815 value for this file is 1,048,576. The value assigned to this
1816 file may be rounded upward, to reflect the value actually
1817 employed for a convenient implementation. To determine the
1818 rounded-up value, display the contents of this file after
1819 assigning a value to it. The minimum value that can be assigned
1820 to this file is the system page size.
1821 /proc/sys/fs/protected_hardlinks (since Linux 3.6)
1823 When the value in this file is 0, no restrictions are placed on
1824 the creation of hard links (i.e., this is the historical behav‐
1825 iour before Linux 3.6). When the value in this file is 1, a
1826 hard link can be created to a target file only if one of the
1827 following conditions is true:
1828 * The caller has the CAP_FOWNER capability.
1830 * The file system UID of the process creating the link matches
1832 the owner (UID) of the target file (as described in creden‐
1833 tials(7), a process's file system UID is normally the same as
1834 its effective UID).
1835 * All of the following conditions are true:
1837 · the target is a regular file;
1839 · the target file does not have its set-user-ID permission
1841 bit enabled;
1842 · the target file does not have both its set-group-ID and
1844 group-executable permission bits enabled; and
1845 · the caller has permission to read and write the target
1847 file (either via the file's permissions mask or because
1848 it has suitable capabilities).
1849 The default value in this file is 0. Setting the value to 1
1851 prevents a longstanding class of security issues caused by hard-
1852 link-based time-of-check, time-of-use races, most commonly seen
1853 in world-writable directories such as /tmp. The common method
1854 of exploiting this flaw is to cross privilege boundaries when
1855 following a given hard link (i.e., a root process follows a hard
1856 link created by another user). Additionally, on systems without
1857 separated partitions, this stops unauthorized users from "pin‐
1858 ning" vulnerable set-user-ID and set-group-ID files against
1859 being upgraded by the administrator, or linking to special
1860 files.
1861 /proc/sys/fs/protected_symlinks (since Linux 3.6)
1863 When the value in this file is 0, no restrictions are placed on
1864 following symbolic links (i.e., this is the historical behaviour
1865 before Linux 3.6). When the value in this file is 1, symbolic
1866 links are followed only in the following circumstances:
1867 * the file system UID of the process following the link matches
1869 the owner (UID) of the symbolic link (as described in creden‐
1870 tials(7), a process's file system UID is normally the same as
1871 its effective UID);
1872 * the link is not in a sticky world-writable directory; or
1874 * the symbolic link and and its parent directory have the same
1876 owner (UID)
1877 A system call that fails to follow a symbolic link because of
1879 the above restrictions returns the error EACCES in errno.
1880 The default value in this file is 0. Setting the value to 1
1882 avoids a longstanding class of security issues based on time-of-
1883 check, time-of-use races when accessing symbolic links.
1884 /proc/sys/fs/suid_dumpable (since Linux 2.6.13)
1886 The value in this file determines whether core dump files are
1887 produced for set-user-ID or otherwise protected/tainted bina‐
1888 ries. Three different integer values can be specified:
1889 0 (default)
1891 This provides the traditional (pre-Linux 2.6.13) behav‐
1892 ior. A core dump will not be produced for a process
1893 which has changed credentials (by calling seteuid(2),
1894 setgid(2), or similar, or by executing a set-user-ID or
1895 set-group-ID program) or whose binary does not have read
1896 permission enabled.
1897 1 ("debug")
1899 All processes dump core when possible. The core dump is
1900 owned by the file system user ID of the dumping process
1901 and no security is applied. This is intended for system
1902 debugging situations only. Ptrace is unchecked.
1903 2 ("suidsafe")
1905 Any binary which normally would not be dumped (see "0"
1906 above) is dumped readable by root only. This allows the
1907 user to remove the core dump file but not to read it.
1908 For security reasons core dumps in this mode will not
1909 overwrite one another or other files. This mode is
1910 appropriate when administrators are attempting to debug
1911 problems in a normal environment.
1912 Additionally, since Linux 3.6, /proc/sys/kernel/core_pat‐
1914 tern must either be an absolute pathname or a pipe com‐
1915 mand, as detailed in core(5). Warnings will be written
1916 to the kernel log if core_pattern does not follow these
1917 rules, and no core dump will be produced.
1918 /proc/sys/fs/super-max
1920 This file controls the maximum number of superblocks, and thus
1921 the maximum number of mounted file systems the kernel can have.
1922 You need increase only super-max if you need to mount more file
1923 systems than the current value in super-max allows you to.
1924 /proc/sys/fs/super-nr
1926 This file contains the number of file systems currently mounted.
1927 /proc/sys/kernel
1929 This directory contains files controlling a range of kernel
1930 parameters, as described below.
1931 /proc/sys/kernel/acct
1933 This file contains three numbers: highwater, lowwater, and fre‐
1934 quency. If BSD-style process accounting is enabled these values
1935 control its behavior. If free space on file system where the
1936 log lives goes below lowwater percent accounting suspends. If
1937 free space gets above highwater percent accounting resumes.
1938 frequency determines how often the kernel checks the amount of
1939 free space (value is in seconds). Default values are 4, 2 and
1940 30. That is, suspend accounting if 2% or less space is free;
1941 resume it if 4% or more space is free; consider information
1942 about amount of free space valid for 30 seconds.
1943 /proc/sys/kernel/cap_last_cap (since Linux 3.2)
1945 See capabilities(7).
1946 /proc/sys/kernel/cap-bound (from Linux 2.2 to 2.6.24)
1948 This file holds the value of the kernel capability bounding set
1949 (expressed as a signed decimal number). This set is ANDed
1950 against the capabilities permitted to a process during
1951 execve(2). Starting with Linux 2.6.25, the system-wide capabil‐
1952 ity bounding set disappeared, and was replaced by a per-thread
1953 bounding set; see capabilities(7).
1954 /proc/sys/kernel/core_pattern
1956 See core(5).
1957 /proc/sys/kernel/core_uses_pid
1959 See core(5).
1960 /proc/sys/kernel/ctrl-alt-del
1962 This file controls the handling of Ctrl-Alt-Del from the key‐
1963 board. When the value in this file is 0, Ctrl-Alt-Del is
1964 trapped and sent to the init(8) program to handle a graceful
1965 restart. When the value is greater than zero, Linux's reaction
1966 to a Vulcan Nerve Pinch (tm) will be an immediate reboot, with‐
1967 out even syncing its dirty buffers. Note: when a program (like
1968 dosemu) has the keyboard in "raw" mode, the ctrl-alt-del is
1969 intercepted by the program before it ever reaches the kernel tty
1970 layer, and it's up to the program to decide what to do with it.
1971 /proc/sys/kernel/dmesg_restrict (since Linux 2.6.37)
1973 The value in this file determines who can see kernel syslog con‐
1974 tents. A value of 0 in this file imposes no restrictions. If
1975 the value is 1, only privileged users can read the kernel sys‐
1976 log. (See syslog(2) for more details.) Since Linux 3.4, only
1977 users with the CAP_SYS_ADMIN capability may change the value in
1978 this file.
1979 /proc/sys/kernel/domainname and /proc/sys/kernel/hostname
1981 can be used to set the NIS/YP domainname and the hostname of
1982 your box in exactly the same way as the commands domainname(1)
1983 and hostname(1), that is:
1984 # echo 'darkstar' > /proc/sys/kernel/hostname
1986 # echo 'mydomain' > /proc/sys/kernel/domainname
1987 has the same effect as
1989 # hostname 'darkstar'
1991 # domainname 'mydomain'
1992 Note, however, that the classic darkstar.frop.org has the host‐
1994 name "darkstar" and DNS (Internet Domain Name Server) domainname
1995 "frop.org", not to be confused with the NIS (Network Information
1996 Service) or YP (Yellow Pages) domainname. These two domain
1997 names are in general different. For a detailed discussion see
1998 the hostname(1) man page.
1999 /proc/sys/kernel/hotplug
2001 This file contains the path for the hotplug policy agent. The
2002 default value in this file is /sbin/hotplug.
2003 /proc/sys/kernel/htab-reclaim
2005 (PowerPC only) If this file is set to a nonzero value, the Pow‐
2006 erPC htab (see kernel file Documentation/powerpc/ppc_htab.txt)
2007 is pruned each time the system hits the idle loop.
2008 /proc/sys/kernel/kptr_restrict (since Linux 2.6.38)
2010 The value in this file determines whether kernel addresses are
2011 exposed via /proc files and other interfaces. A value of 0 in
2012 this file imposes no restrictions. If the value is 1, kernel
2013 pointers printed using the %pK format specifier will be replaced
2014 with zeros unless the user has the CAP_SYSLOG capability. If
2015 the value is 2, kernel pointers printed using the %pK format
2016 specifier will be replaced with zeros regardless of the user's
2017 capabilities. The initial default value for this file was 1,
2018 but the default was changed to 0 in Linux 2.6.39. Since Linux
2019 3.4, only users with the CAP_SYS_ADMIN capability can change the
2020 value in this file.
2021 /proc/sys/kernel/l2cr
2023 (PowerPC only) This file contains a flag that controls the L2
2024 cache of G3 processor boards. If 0, the cache is disabled.
2025 Enabled if nonzero.
2026 /proc/sys/kernel/modprobe
2028 This file contains the path for the kernel module loader. The
2029 default value is /sbin/modprobe. The file is present only if
2030 the kernel is built with the CONFIG_MODULES (CONFIG_KMOD in
2031 Linux 2.6.26 and earlier) option enabled. It is described by
2032 the Linux kernel source file Documentation/kmod.txt (present
2033 only in kernel 2.4 and earlier).
2034 /proc/sys/kernel/modules_disabled (since Linux 2.6.31)
2036 A toggle value indicating if modules are allowed to be loaded in
2037 an otherwise modular kernel. This toggle defaults to off (0),
2038 but can be set true (1). Once true, modules can be neither
2039 loaded nor unloaded, and the toggle cannot be set back to false.
2040 The file is present only if the kernel is built with the CON‐
2041 FIG_MODULES option enabled.
2042 /proc/sys/kernel/msgmax
2044 This file defines a system-wide limit specifying the maximum
2045 number of bytes in a single message written on a System V mes‐
2046 sage queue.
2047 /proc/sys/kernel/msgmni (since Linux 2.4)
2049 This file defines the system-wide limit on the number of message
2050 queue identifiers.
2051 /proc/sys/kernel/msgmnb
2053 This file defines a system-wide parameter used to initialize the
2054 msg_qbytes setting for subsequently created message queues. The
2055 msg_qbytes setting specifies the maximum number of bytes that
2056 may be written to the message queue.
2057 /proc/sys/kernel/ostype and /proc/sys/kernel/osrelease
2059 These files give substrings of /proc/version.
2060 /proc/sys/kernel/overflowgid and /proc/sys/kernel/overflowuid
2062 These files duplicate the files /proc/sys/fs/overflowgid and
2063 /proc/sys/fs/overflowuid.
2064 /proc/sys/kernel/panic
2066 This file gives read/write access to the kernel variable
2067 panic_timeout. If this is zero, the kernel will loop on a
2068 panic; if nonzero it indicates that the kernel should autoreboot
2069 after this number of seconds. When you use the software watch‐
2070 dog device driver, the recommended setting is 60.
2071 /proc/sys/kernel/panic_on_oops (since Linux 2.5.68)
2073 This file controls the kernel's behavior when an oops or BUG is
2074 encountered. If this file contains 0, then the system tries to
2075 continue operation. If it contains 1, then the system delays a
2076 few seconds (to give klogd time to record the oops output) and
2077 then panics. If the /proc/sys/kernel/panic file is also nonzero
2078 then the machine will be rebooted.
2079 /proc/sys/kernel/pid_max (since Linux 2.5.34)
2081 This file specifies the value at which PIDs wrap around (i.e.,
2082 the value in this file is one greater than the maximum PID).
2083 The default value for this file, 32768, results in the same
2084 range of PIDs as on earlier kernels. On 32-bit platforms, 32768
2085 is the maximum value for pid_max. On 64-bit systems, pid_max
2086 can be set to any value up to 2^22 (PID_MAX_LIMIT, approximately
2087 4 million).
2088 /proc/sys/kernel/powersave-nap (PowerPC only)
2090 This file contains a flag. If set, Linux-PPC will use the "nap"
2091 mode of powersaving, otherwise the "doze" mode will be used.
2092 /proc/sys/kernel/printk
2094 The four values in this file are console_loglevel, default_mes‐
2095 sage_loglevel, minimum_console_level, and default_con‐
2096 sole_loglevel. These values influence printk() behavior when
2097 printing or logging error messages. See syslog(2) for more info
2098 on the different loglevels. Messages with a higher priority
2099 than console_loglevel will be printed to the console. Messages
2100 without an explicit priority will be printed with priority
2101 default_message_level. minimum_console_loglevel is the minimum
2102 (highest) value to which console_loglevel can be set.
2103 default_console_loglevel is the default value for con‐
2104 sole_loglevel.
2105 /proc/sys/kernel/pty (since Linux 2.6.4)
2107 This directory contains two files relating to the number of UNIX
2108 98 pseudoterminals (see pts(4)) on the system.
2109 /proc/sys/kernel/pty/max
2111 This file defines the maximum number of pseudoterminals.
2112 /proc/sys/kernel/pty/nr
2114 This read-only file indicates how many pseudoterminals are cur‐
2115 rently in use.
2116 /proc/sys/kernel/random
2118 This directory contains various parameters controlling the oper‐
2119 ation of the file /dev/random. See random(4) for further infor‐
2120 mation.
2121 /proc/sys/kernel/real-root-dev
2123 This file is documented in the Linux kernel source file Documen‐
2124 tation/initrd.txt.
2125 /proc/sys/kernel/reboot-cmd (Sparc only)
2127 This file seems to be a way to give an argument to the SPARC
2128 ROM/Flash boot loader. Maybe to tell it what to do after
2129 rebooting?
2130 /proc/sys/kernel/rtsig-max
2132 (Only in kernels up to and including 2.6.7; see setrlimit(2))
2133 This file can be used to tune the maximum number of POSIX real-
2134 time (queued) signals that can be outstanding in the system.
2135 /proc/sys/kernel/rtsig-nr
2137 (Only in kernels up to and including 2.6.7.) This file shows
2138 the number POSIX real-time signals currently queued.
2139 /proc/sys/kernel/sched_rr_timeslice_ms (since Linux 3.9)
2141 See sched_rr_get_interval(2).
2142 /proc/sys/kernel/sem (since Linux 2.4)
2144 This file contains 4 numbers defining limits for System V IPC
2145 semaphores. These fields are, in order:
2146 SEMMSL The maximum semaphores per semaphore set.
2148 SEMMNS A system-wide limit on the number of semaphores in all
2150 semaphore sets.
2151 SEMOPM The maximum number of operations that may be specified
2153 in a semop(2) call.
2154 SEMMNI A system-wide limit on the maximum number of semaphore
2156 identifiers.
2157 /proc/sys/kernel/sg-big-buff
2159 This file shows the size of the generic SCSI device (sg) buffer.
2160 You can't tune it just yet, but you could change it at compile
2161 time by editing include/scsi/sg.h and changing the value of
2162 SG_BIG_BUFF. However, there shouldn't be any reason to change
2163 this value.
2164 /proc/sys/kernel/shm_rmid_forced (since Linux 3.1)
2166 If this file is set to 1, all System V shared memory segments
2167 will be marked for destruction as soon as the number of attached
2168 processes falls to zero; in other words, it is no longer possi‐
2169 ble to create shared memory segments that exist independently of
2170 any attached process.
2171 The effect is as though a shmctl(2) IPC_RMID is performed on all
2173 existing segments as well as all segments created in the future
2174 (until this file is reset to 0). Note that existing segments
2175 that are attached to no process will be immediately destroyed
2176 when this file is set to 1. Setting this option will also
2177 destroy segments that were created, but never attached, upon
2178 termination of the process that created the segment with
2179 shmget(2).
2180 Setting this file to 1 provides a way of ensuring that all Sys‐
2182 tem V shared memory segments are counted against the resource
2183 usage and resource limits (see the description of RLIMIT_AS in
2184 getrlimit(2)) of at least one process.
2185 Because setting this file to 1 produces behavior that is non‐
2187 standard and could also break existing applications, the default
2188 value in this file is 0. Only set this file to 1 if you have a
2189 good understanding of the semantics of the applications using
2190 System V shared memory on your system.
2191 /proc/sys/kernel/shmall
2193 This file contains the system-wide limit on the total number of
2194 pages of System V shared memory.
2195 /proc/sys/kernel/shmmax
2197 This file can be used to query and set the run-time limit on the
2198 maximum (System V IPC) shared memory segment size that can be
2199 created. Shared memory segments up to 1GB are now supported in
2200 the kernel. This value defaults to SHMMAX.
2201 /proc/sys/kernel/shmmni (since Linux 2.4)
2203 This file specifies the system-wide maximum number of System V
2204 shared memory segments that can be created.
2205 /proc/sys/kernel/sysrq
2207 This file controls the functions allowed to be invoked by the
2208 SysRq key. By default, the file contains 1 meaning that every
2209 possible SysRq request is allowed (in older kernel versions,
2210 SysRq was disabled by default, and you were required to specifi‐
2211 cally enable it at run-time, but this is not the case any more).
2212 Possible values in this file are:
2213 0 - disable sysrq completely
2215 1 - enable all functions of sysrq
2216 >1 - bit mask of allowed sysrq functions, as follows:
2217 2 - enable control of console logging level
2218 4 - enable control of keyboard (SAK, unraw)
2219 8 - enable debugging dumps of processes etc.
2220 16 - enable sync command
2221 32 - enable remount read-only
2222 64 - enable signalling of processes (term, kill, oom-
2223 kill)
2224 128 - allow reboot/poweroff
2225 256 - allow nicing of all real-time tasks
2226 This file is present only if the CONFIG_MAGIC_SYSRQ kernel con‐
2228 figuration option is enabled. For further details see the Linux
2229 kernel source file Documentation/sysrq.txt.
2230 /proc/sys/kernel/version
2232 This file contains a string like:
2233 #5 Wed Feb 25 21:49:24 MET 1998
2235 The "#5" means that this is the fifth kernel built from this
2237 source base and the date behind it indicates the time the kernel
2238 was built.
2239 /proc/sys/kernel/threads-max (since Linux 2.3.11)
2241 This file specifies the system-wide limit on the number of
2242 threads (tasks) that can be created on the system.
2243 /proc/sys/kernel/zero-paged (PowerPC only)
2245 This file contains a flag. When enabled (nonzero), Linux-PPC
2246 will pre-zero pages in the idle loop, possibly speeding up
2247 get_free_pages.
2248 /proc/sys/net
2250 This directory contains networking stuff. Explanations for some
2251 of the files under this directory can be found in tcp(7) and
2252 ip(7).
2253 /proc/sys/net/core/somaxconn
2255 This file defines a ceiling value for the backlog argument of
2256 listen(2); see the listen(2) manual page for details.
2257 /proc/sys/proc
2259 This directory may be empty.
2260 /proc/sys/sunrpc
2262 This directory supports Sun remote procedure call for network
2263 file system (NFS). On some systems, it is not present.
2264 /proc/sys/vm
2266 This directory contains files for memory management tuning, buf‐
2267 fer and cache management.
2268 /proc/sys/vm/drop_caches (since Linux 2.6.16)
2270 Writing to this file causes the kernel to drop clean caches,
2271 dentries and inodes from memory, causing that memory to become
2272 free.
2273 To free pagecache, use echo 1 > /proc/sys/vm/drop_caches; to
2275 free dentries and inodes, use echo 2 > /proc/sys/vm/drop_caches;
2276 to free pagecache, dentries and inodes, use echo 3 >
2277 /proc/sys/vm/drop_caches.
2278 Because this is a nondestructive operation and dirty objects are
2280 not freeable, the user should run sync(8) first.
2281 /proc/sys/vm/legacy_va_layout (since Linux 2.6.9)
2283 If nonzero, this disables the new 32-bit memory-mapping layout;
2284 the kernel will use the legacy (2.4) layout for all processes.
2285 /proc/sys/vm/memory_failure_early_kill (since Linux 2.6.32)
2287 Control how to kill processes when an uncorrected memory error
2288 (typically a 2-bit error in a memory module) that cannot be han‐
2289 dled by the kernel is detected in the background by hardware.
2290 In some cases (like the page still having a valid copy on disk),
2291 the kernel will handle the failure transparently without affect‐
2292 ing any applications. But if there is no other up-to-date copy
2293 of the data, it will kill processes to prevent any data corrup‐
2294 tions from propagating.
2295 The file has one of the following values:
2297 1: Kill all processes that have the corrupted-and-not-reload‐
2299 able page mapped as soon as the corruption is detected.
2300 Note this is not supported for a few types of pages, like
2301 kernel internally allocated data or the swap cache, but
2302 works for the majority of user pages.
2303 0: Only unmap the corrupted page from all processes and kill
2305 only a process that tries to access it.
2306 The kill is performed using a SIGBUS signal with si_code set to
2308 BUS_MCEERR_AO. Processes can handle this if they want to; see
2309 sigaction(2) for more details.
2310 This feature is active only on architectures/platforms with
2312 advanced machine check handling and depends on the hardware
2313 capabilities.
2314 Applications can override the memory_failure_early_kill setting
2316 individually with the prctl(2) PR_MCE_KILL operation.
2317 Only present if the kernel was configured with CONFIG_MEM‐
2319 ORY_FAILURE.
2320 /proc/sys/vm/memory_failure_recovery (since Linux 2.6.32)
2322 Enable memory failure recovery (when supported by the platform)
2323 1: Attempt recovery.
2325 0: Always panic on a memory failure.
2327 Only present if the kernel was configured with CONFIG_MEM‐
2329 ORY_FAILURE.
2330 /proc/sys/vm/oom_dump_tasks (since Linux 2.6.25)
2332 Enables a system-wide task dump (excluding kernel threads) to be
2333 produced when the kernel performs an OOM-killing. The dump
2334 includes the following information for each task (thread,
2335 process): thread ID, real user ID, thread group ID (process ID),
2336 virtual memory size, resident set size, the CPU that the task is
2337 scheduled on, oom_adj score (see the description of
2338 /proc/[pid]/oom_adj), and command name. This is helpful to
2339 determine why the OOM-killer was invoked and to identify the
2340 rogue task that caused it.
2341 If this contains the value zero, this information is suppressed.
2343 On very large systems with thousands of tasks, it may not be
2344 feasible to dump the memory state information for each one.
2345 Such systems should not be forced to incur a performance penalty
2346 in OOM situations when the information may not be desired.
2347 If this is set to nonzero, this information is shown whenever
2349 the OOM-killer actually kills a memory-hogging task.
2350 The default value is 0.
2352 /proc/sys/vm/oom_kill_allocating_task (since Linux 2.6.24)
2354 This enables or disables killing the OOM-triggering task in out-
2355 of-memory situations.
2356 If this is set to zero, the OOM-killer will scan through the
2358 entire tasklist and select a task based on heuristics to kill.
2359 This normally selects a rogue memory-hogging task that frees up
2360 a large amount of memory when killed.
2361 If this is set to nonzero, the OOM-killer simply kills the task
2363 that triggered the out-of-memory condition. This avoids a pos‐
2364 sibly expensive tasklist scan.
2365 If /proc/sys/vm/panic_on_oom is nonzero, it takes precedence
2367 over whatever value is used in /proc/sys/vm/oom_kill_allocat‐
2368 ing_task.
2369 The default value is 0.
2371 /proc/sys/vm/overcommit_memory
2373 This file contains the kernel virtual memory accounting mode.
2374 Values are:
2375 0: heuristic overcommit (this is the default)
2377 1: always overcommit, never check
2378 2: always check, never overcommit
2379 In mode 0, calls of mmap(2) with MAP_NORESERVE are not checked,
2381 and the default check is very weak, leading to the risk of get‐
2382 ting a process "OOM-killed". Under Linux 2.4 any nonzero value
2383 implies mode 1. In mode 2 (available since Linux 2.6), the
2384 total virtual address space on the system is limited to (SS +
2385 RAM*(r/100)), where SS is the size of the swap space, and RAM is
2386 the size of the physical memory, and r is the contents of the
2387 file /proc/sys/vm/overcommit_ratio.
2388 /proc/sys/vm/overcommit_ratio
2390 See the description of /proc/sys/vm/overcommit_memory.
2391 /proc/sys/vm/panic_on_oom (since Linux 2.6.18)
2393 This enables or disables a kernel panic in an out-of-memory sit‐
2394 uation.
2395 If this file is set to the value 0, the kernel's OOM-killer will
2397 kill some rogue process. Usually, the OOM-killer is able to
2398 kill a rogue process and the system will survive.
2399 If this file is set to the value 1, then the kernel normally
2401 panics when out-of-memory happens. However, if a process limits
2402 allocations to certain nodes using memory policies (mbind(2)
2403 MPOL_BIND) or cpusets (cpuset(7)) and those nodes reach memory
2404 exhaustion status, one process may be killed by the OOM-killer.
2405 No panic occurs in this case: because other nodes' memory may be
2406 free, this means the system as a whole may not have reached an
2407 out-of-memory situation yet.
2408 If this file is set to the value 2, the kernel always panics
2410 when an out-of-memory condition occurs.
2411 The default value is 0. 1 and 2 are for failover of clustering.
2413 Select either according to your policy of failover.
2414 /proc/sys/vm/swappiness
2416 The value in this file controls how aggressively the kernel will
2417 swap memory pages. Higher values increase aggressiveness, lower
2418 values decrease aggressiveness. The default value is 60.
2419 /proc/sysrq-trigger (since Linux 2.4.21)
2421 Writing a character to this file triggers the same SysRq func‐
2422 tion as typing ALT-SysRq-<character> (see the description of
2423 /proc/sys/kernel/sysrq). This file is normally writable only by
2424 root. For further details see the Linux kernel source file Doc‐
2425 umentation/sysrq.txt.
2426 /proc/sysvipc
2428 Subdirectory containing the pseudo-files msg, sem and shm.
2429 These files list the System V Interprocess Communication (IPC)
2430 objects (respectively: message queues, semaphores, and shared
2431 memory) that currently exist on the system, providing similar
2432 information to that available via ipcs(1). These files have
2433 headers and are formatted (one IPC object per line) for easy
2434 understanding. svipc(7) provides further background on the
2435 information shown by these files.
2436 /proc/tty
2438 Subdirectory containing the pseudo-files and subdirectories for
2439 tty drivers and line disciplines.
2440 /proc/uptime
2442 This file contains two numbers: the uptime of the system (sec‐
2443 onds), and the amount of time spent in idle process (seconds).
2444 /proc/version
2446 This string identifies the kernel version that is currently run‐
2447 ning. It includes the contents of /proc/sys/kernel/ostype,
2448 /proc/sys/kernel/osrelease and /proc/sys/kernel/version. For
2449 example:
2450 Linux version 1.0.9 (quinlan@phaze) #1 Sat May 14 01:51:54 EDT 1994
2451 /proc/vmstat (since Linux 2.6)
2453 This file displays various virtual memory statistics.
2454 /proc/zoneinfo (since Linux 2.6.13)
2456 This file display information about memory zones. This is use‐
2457 ful for analyzing virtual memory behavior.
2458
2460 Many strings (i.e., the environment and command line) are in the inter‐
2461 nal format, with subfields terminated by null bytes ('\0'), so you may
2462 find that things are more readable if you use od -c or tr "\000" "\n"
2463 to read them. Alternatively, echo `cat <file>` works well.
2464 This manual page is incomplete, possibly inaccurate, and is the kind of
2466 thing that needs to be updated very often.
2467
2469 cat(1), dmesg(1), find(1), free(1), ps(1), tr(1), uptime(1), chroot(2),
2470 mmap(2), readlink(2), syslog(2), slabinfo(5), hier(7), time(7), arp(8),
2471 hdparm(8), ifconfig(8), init(8), lsmod(8), lspci(8), mount(8), net‐
2472 stat(8), procinfo(8), route(8), sysctl(8)
2473 The Linux kernel source files: Documentation/filesystems/proc.txt and
2475 Documentation/sysctl/vm.txt.
2476
2478 This page is part of release 3.53 of the Linux man-pages project. A
2479 description of the project, and information about reporting bugs, can
2480 be found at http://www.kernel.org/doc/man-pages/.
2481Linux 2013-08-01 PROC(5)