1proc(5) File Formats Manual proc(5)
2
6 proc - process information, system information, and sysctl pseudo-
7 filesystem
8
10 The proc filesystem is a pseudo-filesystem which provides an interface
11 to kernel data structures. It is commonly mounted at /proc. Typi‐
12 cally, it is mounted automatically by the system, but it can also be
13 mounted manually using a command such as:
14 mount -t proc proc /proc
16 Most of the files in the proc filesystem are read-only, but some files
18 are writable, allowing kernel variables to be changed.
19 Mount options
21 The proc filesystem supports the following mount options:
22 hidepid=n (since Linux 3.3)
24 This option controls who can access the information in /proc/pid
25 directories. The argument, n, is one of the following values:
26 0 Everybody may access all /proc/pid directories. This is the
28 traditional behavior, and the default if this mount option
29 is not specified.
30 1 Users may not access files and subdirectories inside any
32 /proc/pid directories but their own (the /proc/pid directo‐
33 ries themselves remain visible). Sensitive files such as
34 /proc/pid/cmdline and /proc/pid/status are now protected
35 against other users. This makes it impossible to learn
36 whether any user is running a specific program (so long as
37 the program doesn't otherwise reveal itself by its behav‐
38 ior).
39 2 As for mode 1, but in addition the /proc/pid directories be‐
41 longing to other users become invisible. This means that
42 /proc/pid entries can no longer be used to discover the PIDs
43 on the system. This doesn't hide the fact that a process
44 with a specific PID value exists (it can be learned by other
45 means, for example, by "kill -0 $PID"), but it hides a
46 process's UID and GID, which could otherwise be learned by
47 employing stat(2) on a /proc/pid directory. This greatly
48 complicates an attacker's task of gathering information
49 about running processes (e.g., discovering whether some dae‐
50 mon is running with elevated privileges, whether another
51 user is running some sensitive program, whether other users
52 are running any program at all, and so on).
53 gid=gid (since Linux 3.3)
55 Specifies the ID of a group whose members are authorized to
56 learn process information otherwise prohibited by hidepid (i.e.,
57 users in this group behave as though /proc was mounted with
58 hidepid=0). This group should be used instead of approaches
59 such as putting nonroot users into the sudoers(5) file.
60 Overview
62 Underneath /proc, there are the following general groups of files and
63 subdirectories:
64 /proc/pid subdirectories
66 Each one of these subdirectories contains files and subdirecto‐
67 ries exposing information about the process with the correspond‐
68 ing process ID.
69 Underneath each of the /proc/pid directories, a task subdirec‐
71 tory contains subdirectories of the form task/tid, which contain
72 corresponding information about each of the threads in the
73 process, where tid is the kernel thread ID of the thread.
74 The /proc/pid subdirectories are visible when iterating through
76 /proc with getdents(2) (and thus are visible when one uses ls(1)
77 to view the contents of /proc).
78 /proc/tid subdirectories
80 Each one of these subdirectories contains files and subdirecto‐
81 ries exposing information about the thread with the correspond‐
82 ing thread ID. The contents of these directories are the same
83 as the corresponding /proc/pid/task/tid directories.
84 The /proc/tid subdirectories are not visible when iterating
86 through /proc with getdents(2) (and thus are not visible when
87 one uses ls(1) to view the contents of /proc).
88 /proc/self
90 When a process accesses this magic symbolic link, it resolves to
91 the process's own /proc/pid directory.
92 /proc/thread-self
94 When a thread accesses this magic symbolic link, it resolves to
95 the process's own /proc/self/task/tid directory.
96 /proc/[a-z]*
98 Various other files and subdirectories under /proc expose sys‐
99 tem-wide information.
100 All of the above are described in more detail below.
102 Files and directories
104 The following list provides details of many of the files and directo‐
105 ries under the /proc hierarchy.
106 /proc/pid
108 There is a numerical subdirectory for each running process; the
109 subdirectory is named by the process ID. Each /proc/pid subdi‐
110 rectory contains the pseudo-files and directories described be‐
111 low.
112 The files inside each /proc/pid directory are normally owned by
114 the effective user and effective group ID of the process. How‐
115 ever, as a security measure, the ownership is made root:root if
116 the process's "dumpable" attribute is set to a value other than
117 1.
118 Before Linux 4.11, root:root meant the "global" root user ID and
120 group ID (i.e., UID 0 and GID 0 in the initial user namespace).
121 Since Linux 4.11, if the process is in a noninitial user name‐
122 space that has a valid mapping for user (group) ID 0 inside the
123 namespace, then the user (group) ownership of the files under
124 /proc/pid is instead made the same as the root user (group) ID
125 of the namespace. This means that inside a container, things
126 work as expected for the container "root" user.
127 The process's "dumpable" attribute may change for the following
129 reasons:
130 • The attribute was explicitly set via the prctl(2)
132 PR_SET_DUMPABLE operation.
133 • The attribute was reset to the value in the file
135 /proc/sys/fs/suid_dumpable (described below), for the reasons
136 described in prctl(2).
137 Resetting the "dumpable" attribute to 1 reverts the ownership of
139 the /proc/pid/* files to the process's effective UID and GID.
140 Note, however, that if the effective UID or GID is subsequently
141 modified, then the "dumpable" attribute may be reset, as de‐
142 scribed in prctl(2). Therefore, it may be desirable to reset
143 the "dumpable" attribute after making any desired changes to the
144 process's effective UID or GID.
145 /proc/pid/attr
147 The files in this directory provide an API for security modules.
148 The contents of this directory are files that can be read and
149 written in order to set security-related attributes. This di‐
150 rectory was added to support SELinux, but the intention was that
151 the API be general enough to support other security modules.
152 For the purpose of explanation, examples of how SELinux uses
153 these files are provided below.
154 This directory is present only if the kernel was configured with
156 CONFIG_SECURITY.
157 /proc/pid/attr/current (since Linux 2.6.0)
159 The contents of this file represent the current security at‐
160 tributes of the process.
161 In SELinux, this file is used to get the security context of a
163 process. Prior to Linux 2.6.11, this file could not be used to
164 set the security context (a write was always denied), since
165 SELinux limited process security transitions to execve(2) (see
166 the description of /proc/pid/attr/exec, below). Since Linux
167 2.6.11, SELinux lifted this restriction and began supporting
168 "set" operations via writes to this node if authorized by pol‐
169 icy, although use of this operation is only suitable for appli‐
170 cations that are trusted to maintain any desired separation be‐
171 tween the old and new security contexts.
172 Prior to Linux 2.6.28, SELinux did not allow threads within a
174 multithreaded process to set their security context via this
175 node as it would yield an inconsistency among the security con‐
176 texts of the threads sharing the same memory space. Since Linux
177 2.6.28, SELinux lifted this restriction and began supporting
178 "set" operations for threads within a multithreaded process if
179 the new security context is bounded by the old security context,
180 where the bounded relation is defined in policy and guarantees
181 that the new security context has a subset of the permissions of
182 the old security context.
183 Other security modules may choose to support "set" operations
185 via writes to this node.
186 /proc/pid/attr/exec (since Linux 2.6.0)
188 This file represents the attributes to assign to the process
189 upon a subsequent execve(2).
190 In SELinux, this is needed to support role/domain transitions,
192 and execve(2) is the preferred point to make such transitions
193 because it offers better control over the initialization of the
194 process in the new security label and the inheritance of state.
195 In SELinux, this attribute is reset on execve(2) so that the new
196 program reverts to the default behavior for any execve(2) calls
197 that it may make. In SELinux, a process can set only its own
198 /proc/pid/attr/exec attribute.
199 /proc/pid/attr/fscreate (since Linux 2.6.0)
201 This file represents the attributes to assign to files created
202 by subsequent calls to open(2), mkdir(2), symlink(2), and
203 mknod(2)
204 SELinux employs this file to support creation of a file (using
206 the aforementioned system calls) in a secure state, so that
207 there is no risk of inappropriate access being obtained between
208 the time of creation and the time that attributes are set. In
209 SELinux, this attribute is reset on execve(2), so that the new
210 program reverts to the default behavior for any file creation
211 calls it may make, but the attribute will persist across multi‐
212 ple file creation calls within a program unless it is explicitly
213 reset. In SELinux, a process can set only its own
214 /proc/pid/attr/fscreate attribute.
215 /proc/pid/attr/keycreate (since Linux 2.6.18)
217 If a process writes a security context into this file, all sub‐
218 sequently created keys (add_key(2)) will be labeled with this
219 context. For further information, see the kernel source file
220 Documentation/security/keys/core.rst (or file Documentation/se‐
221 curity/keys.txt between Linux 3.0 and Linux 4.13, or Documenta‐
222 tion/keys.txt before Linux 3.0).
223 /proc/pid/attr/prev (since Linux 2.6.0)
225 This file contains the security context of the process before
226 the last execve(2); that is, the previous value of
227 /proc/pid/attr/current.
228 /proc/pid/attr/socketcreate (since Linux 2.6.18)
230 If a process writes a security context into this file, all sub‐
231 sequently created sockets will be labeled with this context.
232 /proc/pid/autogroup (since Linux 2.6.38)
234 See sched(7).
235 /proc/pid/auxv (since Linux 2.6.0)
237 This contains the contents of the ELF interpreter information
238 passed to the process at exec time. The format is one unsigned
239 long ID plus one unsigned long value for each entry. The last
240 entry contains two zeros. See also getauxval(3).
241 Permission to access this file is governed by a ptrace access
243 mode PTRACE_MODE_READ_FSCREDS check; see ptrace(2).
244 /proc/pid/cgroup (since Linux 2.6.24)
246 See cgroups(7).
247 /proc/pid/clear_refs (since Linux 2.6.22)
249 This is a write-only file, writable only by owner of the
251 process.
252 The following values may be written to the file:
254 1 (since Linux 2.6.22)
256 Reset the PG_Referenced and ACCESSED/YOUNG bits for all
257 the pages associated with the process. (Before Linux
258 2.6.32, writing any nonzero value to this file had this
259 effect.)
260 2 (since Linux 2.6.32)
262 Reset the PG_Referenced and ACCESSED/YOUNG bits for all
263 anonymous pages associated with the process.
264 3 (since Linux 2.6.32)
266 Reset the PG_Referenced and ACCESSED/YOUNG bits for all
267 file-mapped pages associated with the process.
268 Clearing the PG_Referenced and ACCESSED/YOUNG bits provides a
270 method to measure approximately how much memory a process is us‐
271 ing. One first inspects the values in the "Referenced" fields
272 for the VMAs shown in /proc/pid/smaps to get an idea of the mem‐
273 ory footprint of the process. One then clears the PG_Referenced
274 and ACCESSED/YOUNG bits and, after some measured time interval,
275 once again inspects the values in the "Referenced" fields to get
276 an idea of the change in memory footprint of the process during
277 the measured interval. If one is interested only in inspecting
278 the selected mapping types, then the value 2 or 3 can be used
279 instead of 1.
280 Further values can be written to affect different properties:
282 4 (since Linux 3.11)
284 Clear the soft-dirty bit for all the pages associated
285 with the process. This is used (in conjunction with
286 /proc/pid/pagemap) by the check-point restore system to
287 discover which pages of a process have been dirtied since
288 the file /proc/pid/clear_refs was written to.
289 5 (since Linux 4.0)
291 Reset the peak resident set size ("high water mark") to
292 the process's current resident set size value.
293 Writing any value to /proc/pid/clear_refs other than those
295 listed above has no effect.
296 The /proc/pid/clear_refs file is present only if the CON‐
298 FIG_PROC_PAGE_MONITOR kernel configuration option is enabled.
299 /proc/pid/cmdline
301 This read-only file holds the complete command line for the
302 process, unless the process is a zombie. In the latter case,
303 there is nothing in this file: that is, a read on this file will
304 return 0 characters. The command-line arguments appear in this
305 file as a set of strings separated by null bytes ('\0'), with a
306 further null byte after the last string.
307 If, after an execve(2), the process modifies its argv strings,
309 those changes will show up here. This is not the same thing as
310 modifying the argv array.
311 Furthermore, a process may change the memory location that this
313 file refers via prctl(2) operations such as PR_SET_MM_ARG_START.
314 Think of this file as the command line that the process wants
316 you to see.
317 /proc/pid/comm (since Linux 2.6.33)
319 This file exposes the process's comm value—that is, the command
320 name associated with the process. Different threads in the same
321 process may have different comm values, accessible via
322 /proc/pid/task/tid/comm. A thread may modify its comm value, or
323 that of any of other thread in the same thread group (see the
324 discussion of CLONE_THREAD in clone(2)), by writing to the file
325 /proc/self/task/tid/comm. Strings longer than TASK_COMM_LEN
326 (16) characters (including the terminating null byte) are
327 silently truncated.
328 This file provides a superset of the prctl(2) PR_SET_NAME and
330 PR_GET_NAME operations, and is employed by pthread_setname_np(3)
331 when used to rename threads other than the caller. The value in
332 this file is used for the %e specifier in /proc/sys/ker‐
333 nel/core_pattern; see core(5).
334 /proc/pid/coredump_filter (since Linux 2.6.23)
336 See core(5).
337 /proc/pid/cpuset (since Linux 2.6.12)
339 See cpuset(7).
340 /proc/pid/cwd
342 This is a symbolic link to the current working directory of the
343 process. To find out the current working directory of process
344 20, for instance, you can do this:
345 $ cd /proc/20/cwd; pwd -P
347 In a multithreaded process, the contents of this symbolic link
349 are not available if the main thread has already terminated
350 (typically by calling pthread_exit(3)).
351 Permission to dereference or read (readlink(2)) this symbolic
353 link is governed by a ptrace access mode PTRACE_MODE_READ_FS‐
354 CREDS check; see ptrace(2).
355 /proc/pid/environ
357 This file contains the initial environment that was set when the
358 currently executing program was started via execve(2). The en‐
359 tries are separated by null bytes ('\0'), and there may be a
360 null byte at the end. Thus, to print out the environment of
361 process 1, you would do:
362 $ cat /proc/1/environ | tr '\000' '\n'
364 If, after an execve(2), the process modifies its environment
366 (e.g., by calling functions such as putenv(3) or modifying the
367 environ(7) variable directly), this file will not reflect those
368 changes.
369 Furthermore, a process may change the memory location that this
371 file refers via prctl(2) operations such as PR_SET_MM_ENV_START.
372 Permission to access this file is governed by a ptrace access
374 mode PTRACE_MODE_READ_FSCREDS check; see ptrace(2).
375 /proc/pid/exe
377 Under Linux 2.2 and later, this file is a symbolic link contain‐
378 ing the actual pathname of the executed command. This symbolic
379 link can be dereferenced normally; attempting to open it will
380 open the executable. You can even type /proc/pid/exe to run an‐
381 other copy of the same executable that is being run by process
382 pid. If the pathname has been unlinked, the symbolic link will
383 contain the string '(deleted)' appended to the original path‐
384 name. In a multithreaded process, the contents of this symbolic
385 link are not available if the main thread has already terminated
386 (typically by calling pthread_exit(3)).
387 Permission to dereference or read (readlink(2)) this symbolic
389 link is governed by a ptrace access mode PTRACE_MODE_READ_FS‐
390 CREDS check; see ptrace(2).
391 Under Linux 2.0 and earlier, /proc/pid/exe is a pointer to the
393 binary which was executed, and appears as a symbolic link. A
394 readlink(2) call on this file under Linux 2.0 returns a string
395 in the format:
396 [device]:inode
398 For example, [0301]:1502 would be inode 1502 on device major 03
400 (IDE, MFM, etc. drives) minor 01 (first partition on the first
401 drive).
402 find(1) with the -inum option can be used to locate the file.
404 /proc/pid/fd/
406 This is a subdirectory containing one entry for each file which
407 the process has open, named by its file descriptor, and which is
408 a symbolic link to the actual file. Thus, 0 is standard input,
409 1 standard output, 2 standard error, and so on.
410 For file descriptors for pipes and sockets, the entries will be
412 symbolic links whose content is the file type with the inode. A
413 readlink(2) call on this file returns a string in the format:
414 type:[inode]
416 For example, socket:[2248868] will be a socket and its inode is
418 2248868. For sockets, that inode can be used to find more in‐
419 formation in one of the files under /proc/net/.
420 For file descriptors that have no corresponding inode (e.g.,
422 file descriptors produced by bpf(2), epoll_create(2),
423 eventfd(2), inotify_init(2), perf_event_open(2), signalfd(2),
424 timerfd_create(2), and userfaultfd(2)), the entry will be a sym‐
425 bolic link with contents of the form
426 anon_inode:file-type
428 In many cases (but not all), the file-type is surrounded by
430 square brackets.
431 For example, an epoll file descriptor will have a symbolic link
433 whose content is the string anon_inode:[eventpoll].
434 In a multithreaded process, the contents of this directory are
436 not available if the main thread has already terminated (typi‐
437 cally by calling pthread_exit(3)).
438 Programs that take a filename as a command-line argument, but
440 don't take input from standard input if no argument is supplied,
441 and programs that write to a file named as a command-line argu‐
442 ment, but don't send their output to standard output if no argu‐
443 ment is supplied, can nevertheless be made to use standard input
444 or standard output by using /proc/pid/fd files as command-line
445 arguments. For example, assuming that -i is the flag designat‐
446 ing an input file and -o is the flag designating an output file:
447 $ foobar -i /proc/self/fd/0 -o /proc/self/fd/1 ...
449 and you have a working filter.
451 /proc/self/fd/N is approximately the same as /dev/fd/N in some
453 UNIX and UNIX-like systems. Most Linux MAKEDEV scripts symboli‐
454 cally link /dev/fd to /proc/self/fd, in fact.
455 Most systems provide symbolic links /dev/stdin, /dev/stdout, and
457 /dev/stderr, which respectively link to the files 0, 1, and 2 in
458 /proc/self/fd. Thus the example command above could be written
459 as:
460 $ foobar -i /dev/stdin -o /dev/stdout ...
462 Permission to dereference or read (readlink(2)) the symbolic
464 links in this directory is governed by a ptrace access mode
465 PTRACE_MODE_READ_FSCREDS check; see ptrace(2).
466 Note that for file descriptors referring to inodes (pipes and
468 sockets, see above), those inodes still have permission bits and
469 ownership information distinct from those of the /proc/pid/fd
470 entry, and that the owner may differ from the user and group IDs
471 of the process. An unprivileged process may lack permissions to
472 open them, as in this example:
473 $ echo test | sudo -u nobody cat
475 test
476 $ echo test | sudo -u nobody cat /proc/self/fd/0
477 cat: /proc/self/fd/0: Permission denied
478 File descriptor 0 refers to the pipe created by the shell and
480 owned by that shell's user, which is not nobody, so cat does not
481 have permission to create a new file descriptor to read from
482 that inode, even though it can still read from its existing file
483 descriptor 0.
484 /proc/pid/fdinfo/ (since Linux 2.6.22)
486 This is a subdirectory containing one entry for each file which
487 the process has open, named by its file descriptor. The files
488 in this directory are readable only by the owner of the process.
489 The contents of each file can be read to obtain information
490 about the corresponding file descriptor. The content depends on
491 the type of file referred to by the corresponding file descrip‐
492 tor.
493 For regular files and directories, we see something like:
495 $ cat /proc/12015/fdinfo/4
497 pos: 1000
498 flags: 01002002
499 mnt_id: 21
500 The fields are as follows:
502 pos This is a decimal number showing the file offset.
504 flags This is an octal number that displays the file access
506 mode and file status flags (see open(2)). If the close-
507 on-exec file descriptor flag is set, then flags will also
508 include the value O_CLOEXEC.
509 Before Linux 3.1, this field incorrectly displayed the
511 setting of O_CLOEXEC at the time the file was opened,
512 rather than the current setting of the close-on-exec
513 flag.
514 mnt_id This field, present since Linux 3.15, is the ID of the
516 mount containing this file. See the description of
517 /proc/pid/mountinfo.
518 For eventfd file descriptors (see eventfd(2)), we see (since
520 Linux 3.8) the following fields:
521 pos: 0
523 flags: 02
524 mnt_id: 10
525 eventfd-count: 40
526 eventfd-count is the current value of the eventfd counter, in
528 hexadecimal.
529 For epoll file descriptors (see epoll(7)), we see (since Linux
531 3.8) the following fields:
532 pos: 0
534 flags: 02
535 mnt_id: 10
536 tfd: 9 events: 19 data: 74253d2500000009
537 tfd: 7 events: 19 data: 74253d2500000007
538 Each of the lines beginning tfd describes one of the file de‐
540 scriptors being monitored via the epoll file descriptor (see
541 epoll_ctl(2) for some details). The tfd field is the number of
542 the file descriptor. The events field is a hexadecimal mask of
543 the events being monitored for this file descriptor. The data
544 field is the data value associated with this file descriptor.
545 For signalfd file descriptors (see signalfd(2)), we see (since
547 Linux 3.8) the following fields:
548 pos: 0
550 flags: 02
551 mnt_id: 10
552 sigmask: 0000000000000006
553 sigmask is the hexadecimal mask of signals that are accepted via
555 this signalfd file descriptor. (In this example, bits 2 and 3
556 are set, corresponding to the signals SIGINT and SIGQUIT; see
557 signal(7).)
558 For inotify file descriptors (see inotify(7)), we see (since
560 Linux 3.8) the following fields:
561 pos: 0
563 flags: 00
564 mnt_id: 11
565 inotify wd:2 ino:7ef82a sdev:800001 mask:800afff ignored_mask:0 fhandle-bytes:8 fhandle-type:1 f_handle:2af87e00220ffd73
566 inotify wd:1 ino:192627 sdev:800001 mask:800afff ignored_mask:0 fhandle-bytes:8 fhandle-type:1 f_handle:27261900802dfd73
567 Each of the lines beginning with "inotify" displays information
569 about one file or directory that is being monitored. The fields
570 in this line are as follows:
571 wd A watch descriptor number (in decimal).
573 ino The inode number of the target file (in hexadecimal).
575 sdev The ID of the device where the target file resides (in
577 hexadecimal).
578 mask The mask of events being monitored for the target file
580 (in hexadecimal).
581 If the kernel was built with exportfs support, the path to the
583 target file is exposed as a file handle, via three hexadecimal
584 fields: fhandle-bytes, fhandle-type, and f_handle.
585 For fanotify file descriptors (see fanotify(7)), we see (since
587 Linux 3.8) the following fields:
588 pos: 0
590 flags: 02
591 mnt_id: 11
592 fanotify flags:0 event-flags:88002
593 fanotify ino:19264f sdev:800001 mflags:0 mask:1 ignored_mask:0 fhandle-bytes:8 fhandle-type:1 f_handle:4f261900a82dfd73
594 The fourth line displays information defined when the fanotify
596 group was created via fanotify_init(2):
597 flags The flags argument given to fanotify_init(2) (expressed
599 in hexadecimal).
600 event-flags
602 The event_f_flags argument given to fanotify_init(2) (ex‐
603 pressed in hexadecimal).
604 Each additional line shown in the file contains information
606 about one of the marks in the fanotify group. Most of these
607 fields are as for inotify, except:
608 mflags The flags associated with the mark (expressed in hexadec‐
610 imal).
611 mask The events mask for this mark (expressed in hexadecimal).
613 ignored_mask
615 The mask of events that are ignored for this mark (ex‐
616 pressed in hexadecimal).
617 For details on these fields, see fanotify_mark(2).
619 For timerfd file descriptors (see timerfd(2)), we see (since
621 Linux 3.17) the following fields:
622 pos: 0
624 flags: 02004002
625 mnt_id: 13
626 clockid: 0
627 ticks: 0
628 settime flags: 03
629 it_value: (7695568592, 640020877)
630 it_interval: (0, 0)
631 clockid
633 This is the numeric value of the clock ID (corresponding
634 to one of the CLOCK_* constants defined via <time.h>)
635 that is used to mark the progress of the timer (in this
636 example, 0 is CLOCK_REALTIME).
637 ticks This is the number of timer expirations that have oc‐
639 curred, (i.e., the value that read(2) on it would re‐
640 turn).
641 settime flags
643 This field lists the flags with which the timerfd was
644 last armed (see timerfd_settime(2)), in octal (in this
645 example, both TFD_TIMER_ABSTIME and TFD_TIMER_CAN‐
646 CEL_ON_SET are set).
647 it_value
649 This field contains the amount of time until the timer
650 will next expire, expressed in seconds and nanoseconds.
651 This is always expressed as a relative value, regardless
652 of whether the timer was created using the TFD_TIMER_AB‐
653 STIME flag.
654 it_interval
656 This field contains the interval of the timer, in seconds
657 and nanoseconds. (The it_value and it_interval fields
658 contain the values that timerfd_gettime(2) on this file
659 descriptor would return.)
660 /proc/pid/gid_map (since Linux 3.5)
662 See user_namespaces(7).
663 /proc/pid/io (since Linux 2.6.20)
665 This file contains I/O statistics for the process, for example:
666 # cat /proc/3828/io
668 rchar: 323934931
669 wchar: 323929600
670 syscr: 632687
671 syscw: 632675
672 read_bytes: 0
673 write_bytes: 323932160
674 cancelled_write_bytes: 0
675 The fields are as follows:
677 rchar: characters read
679 The number of bytes which this task has caused to be read
680 from storage. This is simply the sum of bytes which this
681 process passed to read(2) and similar system calls. It
682 includes things such as terminal I/O and is unaffected by
683 whether or not actual physical disk I/O was required (the
684 read might have been satisfied from pagecache).
685 wchar: characters written
687 The number of bytes which this task has caused, or shall
688 cause to be written to disk. Similar caveats apply here
689 as with rchar.
690 syscr: read syscalls
692 Attempt to count the number of read I/O operations—that
693 is, system calls such as read(2) and pread(2).
694 syscw: write syscalls
696 Attempt to count the number of write I/O operations—that
697 is, system calls such as write(2) and pwrite(2).
698 read_bytes: bytes read
700 Attempt to count the number of bytes which this process
701 really did cause to be fetched from the storage layer.
702 This is accurate for block-backed filesystems.
703 write_bytes: bytes written
705 Attempt to count the number of bytes which this process
706 caused to be sent to the storage layer.
707 cancelled_write_bytes:
709 The big inaccuracy here is truncate. If a process writes
710 1 MB to a file and then deletes the file, it will in fact
711 perform no writeout. But it will have been accounted as
712 having caused 1 MB of write. In other words: this field
713 represents the number of bytes which this process caused
714 to not happen, by truncating pagecache. A task can cause
715 "negative" I/O too. If this task truncates some dirty
716 pagecache, some I/O which another task has been accounted
717 for (in its write_bytes) will not be happening.
718 Note: In the current implementation, things are a bit racy on
720 32-bit systems: if process A reads process B's /proc/pid/io
721 while process B is updating one of these 64-bit counters,
722 process A could see an intermediate result.
723 Permission to access this file is governed by a ptrace access
725 mode PTRACE_MODE_READ_FSCREDS check; see ptrace(2).
726 /proc/pid/limits (since Linux 2.6.24)
728 This file displays the soft limit, hard limit, and units of mea‐
729 surement for each of the process's resource limits (see getr‐
730 limit(2)). Up to and including Linux 2.6.35, this file is pro‐
731 tected to allow reading only by the real UID of the process.
732 Since Linux 2.6.36, this file is readable by all users on the
733 system.
734 /proc/pid/map_files/ (since Linux 3.3)
736 This subdirectory contains entries corresponding to memory-
737 mapped files (see mmap(2)). Entries are named by memory region
738 start and end address pair (expressed as hexadecimal numbers),
739 and are symbolic links to the mapped files themselves. Here is
740 an example, with the output wrapped and reformatted to fit on an
741 80-column display:
742 # ls -l /proc/self/map_files/
744 lr--------. 1 root root 64 Apr 16 21:31
745 3252e00000-3252e20000 -> /usr/lib64/ld-2.15.so
746 ...
747 Although these entries are present for memory regions that were
749 mapped with the MAP_FILE flag, the way anonymous shared memory
750 (regions created with the MAP_ANON | MAP_SHARED flags) is imple‐
751 mented in Linux means that such regions also appear on this di‐
752 rectory. Here is an example where the target file is the
753 deleted /dev/zero one:
754 lrw-------. 1 root root 64 Apr 16 21:33
756 7fc075d2f000-7fc075e6f000 -> /dev/zero (deleted)
757 Permission to access this file is governed by a ptrace access
759 mode PTRACE_MODE_READ_FSCREDS check; see ptrace(2).
760 Until Linux 4.3, this directory appeared only if the CON‐
762 FIG_CHECKPOINT_RESTORE kernel configuration option was enabled.
763 Capabilities are required to read the contents of the symbolic
765 links in this directory: before Linux 5.9, the reading process
766 requires CAP_SYS_ADMIN in the initial user namespace; since
767 Linux 5.9, the reading process must have either CAP_SYS_ADMIN or
768 CAP_CHECKPOINT_RESTORE in the initial (i.e. root) user name‐
769 space.
770 /proc/pid/maps
772 A file containing the currently mapped memory regions and their
773 access permissions. See mmap(2) for some further information
774 about memory mappings.
775 Permission to access this file is governed by a ptrace access
777 mode PTRACE_MODE_READ_FSCREDS check; see ptrace(2).
778 The format of the file is:
780 address perms offset dev inode pathname
782 00400000-00452000 r-xp 00000000 08:02 173521 /usr/bin/dbus-daemon
783 00651000-00652000 r--p 00051000 08:02 173521 /usr/bin/dbus-daemon
784 00652000-00655000 rw-p 00052000 08:02 173521 /usr/bin/dbus-daemon
785 00e03000-00e24000 rw-p 00000000 00:00 0 [heap]
786 00e24000-011f7000 rw-p 00000000 00:00 0 [heap]
787 ...
788 35b1800000-35b1820000 r-xp 00000000 08:02 135522 /usr/lib64/ld-2.15.so
789 35b1a1f000-35b1a20000 r--p 0001f000 08:02 135522 /usr/lib64/ld-2.15.so
790 35b1a20000-35b1a21000 rw-p 00020000 08:02 135522 /usr/lib64/ld-2.15.so
791 35b1a21000-35b1a22000 rw-p 00000000 00:00 0
792 35b1c00000-35b1dac000 r-xp 00000000 08:02 135870 /usr/lib64/libc-2.15.so
793 35b1dac000-35b1fac000 ---p 001ac000 08:02 135870 /usr/lib64/libc-2.15.so
794 35b1fac000-35b1fb0000 r--p 001ac000 08:02 135870 /usr/lib64/libc-2.15.so
795 35b1fb0000-35b1fb2000 rw-p 001b0000 08:02 135870 /usr/lib64/libc-2.15.so
796 ...
797 f2c6ff8c000-7f2c7078c000 rw-p 00000000 00:00 0 [stack:986]
798 ...
799 7fffb2c0d000-7fffb2c2e000 rw-p 00000000 00:00 0 [stack]
800 7fffb2d48000-7fffb2d49000 r-xp 00000000 00:00 0 [vdso]
801 The address field is the address space in the process that the
803 mapping occupies. The perms field is a set of permissions:
804 r = read
806 w = write
807 x = execute
808 s = shared
809 p = private (copy on write)
810 The offset field is the offset into the file/whatever; dev is
812 the device (major:minor); inode is the inode on that device. 0
813 indicates that no inode is associated with the memory region, as
814 would be the case with BSS (uninitialized data).
815 The pathname field will usually be the file that is backing the
817 mapping. For ELF files, you can easily coordinate with the off‐
818 set field by looking at the Offset field in the ELF program
819 headers (readelf -l).
820 There are additional helpful pseudo-paths:
822 [stack]
824 The initial process's (also known as the main thread's)
825 stack.
826 [stack:tid] (from Linux 3.4 to Linux 4.4)
828 A thread's stack (where the tid is a thread ID). It cor‐
829 responds to the /proc/pid/task/tid/ path. This field was
830 removed in Linux 4.5, since providing this information
831 for a process with large numbers of threads is expensive.
832 [vdso] The virtual dynamically linked shared object. See
834 vdso(7).
835 [heap] The process's heap.
837 [anon:name] (since Linux 5.17)
839 A named private anonymous mapping. Set with prctl(2)
840 PR_SET_VMA_ANON_NAME.
841 [anon_shmem:name] (since Linux 6.2)
843 A named shared anonymous mapping. Set with prctl(2)
844 PR_SET_VMA_ANON_NAME.
845 If the pathname field is blank, this is an anonymous mapping as
847 obtained via mmap(2). There is no easy way to coordinate this
848 back to a process's source, short of running it through gdb(1),
849 strace(1), or similar.
850 pathname is shown unescaped except for newline characters, which
852 are replaced with an octal escape sequence. As a result, it is
853 not possible to determine whether the original pathname con‐
854 tained a newline character or the literal \012 character se‐
855 quence.
856 If the mapping is file-backed and the file has been deleted, the
858 string " (deleted)" is appended to the pathname. Note that this
859 is ambiguous too.
860 Under Linux 2.0, there is no field giving pathname.
862 /proc/pid/mem
864 This file can be used to access the pages of a process's memory
865 through open(2), read(2), and lseek(2).
866 Permission to access this file is governed by a ptrace access
868 mode PTRACE_MODE_ATTACH_FSCREDS check; see ptrace(2).
869 /proc/pid/mountinfo (since Linux 2.6.26)
871 This file contains information about mounts in the process's
872 mount namespace (see mount_namespaces(7)). It supplies various
873 information (e.g., propagation state, root of mount for bind
874 mounts, identifier for each mount and its parent) that is miss‐
875 ing from the (older) /proc/pid/mounts file, and fixes various
876 other problems with that file (e.g., nonextensibility, failure
877 to distinguish per-mount versus per-superblock options).
878 The file contains lines of the form:
880 36 35 98:0 /mnt1 /mnt2 rw,noatime master:1 - ext3 /dev/root rw,errors=continue
882 (1)(2)(3) (4) (5) (6) (7) (8) (9) (10) (11)
883 The numbers in parentheses are labels for the descriptions be‐
885 low:
886 (1) mount ID: a unique ID for the mount (may be reused after
888 umount(2)).
889 (2) parent ID: the ID of the parent mount (or of self for the
891 root of this mount namespace's mount tree).
892 If a new mount is stacked on top of a previous existing
894 mount (so that it hides the existing mount) at pathname P,
895 then the parent of the new mount is the previous mount at
896 that location. Thus, when looking at all the mounts
897 stacked at a particular location, the top-most mount is the
898 one that is not the parent of any other mount at the same
899 location. (Note, however, that this top-most mount will be
900 accessible only if the longest path subprefix of P that is
901 a mount point is not itself hidden by a stacked mount.)
902 If the parent mount lies outside the process's root direc‐
904 tory (see chroot(2)), the ID shown here won't have a corre‐
905 sponding record in mountinfo whose mount ID (field 1)
906 matches this parent mount ID (because mounts that lie out‐
907 side the process's root directory are not shown in
908 mountinfo). As a special case of this point, the process's
909 root mount may have a parent mount (for the initramfs
910 filesystem) that lies outside the process's root directory,
911 and an entry for that mount will not appear in mountinfo.
912 (3) major:minor: the value of st_dev for files on this filesys‐
914 tem (see stat(2)).
915 (4) root: the pathname of the directory in the filesystem which
917 forms the root of this mount.
918 (5) mount point: the pathname of the mount point relative to
920 the process's root directory.
921 (6) mount options: per-mount options (see mount(2)).
923 (7) optional fields: zero or more fields of the form
925 "tag[:value]"; see below.
926 (8) separator: the end of the optional fields is marked by a
928 single hyphen.
929 (9) filesystem type: the filesystem type in the form
931 "type[.subtype]".
932 (10) mount source: filesystem-specific information or "none".
934 (11) super options: per-superblock options (see mount(2)).
936 Currently, the possible optional fields are shared, master,
938 propagate_from, and unbindable. See mount_namespaces(7) for a
939 description of these fields. Parsers should ignore all unrecog‐
940 nized optional fields.
941 For more information on mount propagation see Documenta‐
943 tion/filesystems/sharedsubtree.rst (or Documentation/filesys‐
944 tems/sharedsubtree.txt before Linux 5.8) in the Linux kernel
945 source tree.
946 /proc/pid/mounts (since Linux 2.4.19)
948 This file lists all the filesystems currently mounted in the
949 process's mount namespace (see mount_namespaces(7)). The format
950 of this file is documented in fstab(5).
951 Since Linux 2.6.15, this file is pollable: after opening the
953 file for reading, a change in this file (i.e., a filesystem
954 mount or unmount) causes select(2) to mark the file descriptor
955 as having an exceptional condition, and poll(2) and
956 epoll_wait(2) mark the file as having a priority event (POLL‐
957 PRI). (Before Linux 2.6.30, a change in this file was indicated
958 by the file descriptor being marked as readable for select(2),
959 and being marked as having an error condition for poll(2) and
960 epoll_wait(2).)
961 /proc/pid/mountstats (since Linux 2.6.17)
963 This file exports information (statistics, configuration infor‐
964 mation) about the mounts in the process's mount namespace (see
965 mount_namespaces(7)). Lines in this file have the form:
966 device /dev/sda7 mounted on /home with fstype ext3 [stats]
968 ( 1 ) ( 2 ) (3 ) ( 4 )
969 The fields in each line are:
971 (1) The name of the mounted device (or "nodevice" if there is
973 no corresponding device).
974 (2) The mount point within the filesystem tree.
976 (3) The filesystem type.
978 (4) Optional statistics and configuration information. Cur‐
980 rently (as at Linux 2.6.26), only NFS filesystems export
981 information via this field.
982 This file is readable only by the owner of the process.
984 /proc/pid/net (since Linux 2.6.25)
986 See the description of /proc/net.
987 /proc/pid/ns/ (since Linux 3.0)
989 This is a subdirectory containing one entry for each namespace
990 that supports being manipulated by setns(2). For more informa‐
991 tion, see namespaces(7).
992 /proc/pid/numa_maps (since Linux 2.6.14)
994 See numa(7).
995 /proc/pid/oom_adj (since Linux 2.6.11)
997 This file can be used to adjust the score used to select which
998 process should be killed in an out-of-memory (OOM) situation.
999 The kernel uses this value for a bit-shift operation of the
1000 process's oom_score value: valid values are in the range -16 to
1001 +15, plus the special value -17, which disables OOM-killing al‐
1002 together for this process. A positive score increases the like‐
1003 lihood of this process being killed by the OOM-killer; a nega‐
1004 tive score decreases the likelihood.
1005 The default value for this file is 0; a new process inherits its
1007 parent's oom_adj setting. A process must be privileged
1008 (CAP_SYS_RESOURCE) to update this file.
1009 Since Linux 2.6.36, use of this file is deprecated in favor of
1011 /proc/pid/oom_score_adj.
1012 /proc/pid/oom_score (since Linux 2.6.11)
1014 This file displays the current score that the kernel gives to
1015 this process for the purpose of selecting a process for the OOM-
1016 killer. A higher score means that the process is more likely to
1017 be selected by the OOM-killer. The basis for this score is the
1018 amount of memory used by the process, with increases (+) or de‐
1019 creases (-) for factors including:
1020 • whether the process is privileged (-).
1022 Before Linux 2.6.36 the following factors were also used in the
1024 calculation of oom_score:
1025 • whether the process creates a lot of children using fork(2)
1027 (+);
1028 • whether the process has been running a long time, or has used
1030 a lot of CPU time (-);
1031 • whether the process has a low nice value (i.e., > 0) (+); and
1033 • whether the process is making direct hardware access (-).
1035 The oom_score also reflects the adjustment specified by the
1037 oom_score_adj or oom_adj setting for the process.
1038 /proc/pid/oom_score_adj (since Linux 2.6.36)
1040 This file can be used to adjust the badness heuristic used to
1041 select which process gets killed in out-of-memory conditions.
1042 The badness heuristic assigns a value to each candidate task
1044 ranging from 0 (never kill) to 1000 (always kill) to determine
1045 which process is targeted. The units are roughly a proportion
1046 along that range of allowed memory the process may allocate
1047 from, based on an estimation of its current memory and swap use.
1048 For example, if a task is using all allowed memory, its badness
1049 score will be 1000. If it is using half of its allowed memory,
1050 its score will be 500.
1051 There is an additional factor included in the badness score:
1053 root processes are given 3% extra memory over other tasks.
1054 The amount of "allowed" memory depends on the context in which
1056 the OOM-killer was called. If it is due to the memory assigned
1057 to the allocating task's cpuset being exhausted, the allowed
1058 memory represents the set of mems assigned to that cpuset (see
1059 cpuset(7)). If it is due to a mempolicy's node(s) being ex‐
1060 hausted, the allowed memory represents the set of mempolicy
1061 nodes. If it is due to a memory limit (or swap limit) being
1062 reached, the allowed memory is that configured limit. Finally,
1063 if it is due to the entire system being out of memory, the al‐
1064 lowed memory represents all allocatable resources.
1065 The value of oom_score_adj is added to the badness score before
1067 it is used to determine which task to kill. Acceptable values
1068 range from -1000 (OOM_SCORE_ADJ_MIN) to +1000
1069 (OOM_SCORE_ADJ_MAX). This allows user space to control the
1070 preference for OOM-killing, ranging from always preferring a
1071 certain task or completely disabling it from OOM-killing. The
1072 lowest possible value, -1000, is equivalent to disabling OOM-
1073 killing entirely for that task, since it will always report a
1074 badness score of 0.
1075 Consequently, it is very simple for user space to define the
1077 amount of memory to consider for each task. Setting an
1078 oom_score_adj value of +500, for example, is roughly equivalent
1079 to allowing the remainder of tasks sharing the same system,
1080 cpuset, mempolicy, or memory controller resources to use at
1081 least 50% more memory. A value of -500, on the other hand,
1082 would be roughly equivalent to discounting 50% of the task's al‐
1083 lowed memory from being considered as scoring against the task.
1084 For backward compatibility with previous kernels,
1086 /proc/pid/oom_adj can still be used to tune the badness score.
1087 Its value is scaled linearly with oom_score_adj.
1088 Writing to /proc/pid/oom_score_adj or /proc/pid/oom_adj will
1090 change the other with its scaled value.
1091 The choom(1) program provides a command-line interface for ad‐
1093 justing the oom_score_adj value of a running process or a newly
1094 executed command.
1095 /proc/pid/pagemap (since Linux 2.6.25)
1097 This file shows the mapping of each of the process's virtual
1098 pages into physical page frames or swap area. It contains one
1099 64-bit value for each virtual page, with the bits set as fol‐
1100 lows:
1101 63 If set, the page is present in RAM.
1103 62 If set, the page is in swap space
1105 61 (since Linux 3.5)
1107 The page is a file-mapped page or a shared anonymous
1108 page.
1109 60–58 (since Linux 3.11)
1111 Zero
1112 57 (since Linux 5.14)
1114 If set, the page is write-protected through user‐
1115 faultfd(2).
1116 56 (since Linux 4.2)
1118 The page is exclusively mapped.
1119 55 (since Linux 3.11)
1121 PTE is soft-dirty (see the kernel source file Documenta‐
1122 tion/admin-guide/mm/soft-dirty.rst).
1123 54–0 If the page is present in RAM (bit 63), then these bits
1125 provide the page frame number, which can be used to index
1126 /proc/kpageflags and /proc/kpagecount. If the page is
1127 present in swap (bit 62), then bits 4–0 give the swap
1128 type, and bits 54–5 encode the swap offset.
1129 Before Linux 3.11, bits 60–55 were used to encode the base-2 log
1131 of the page size.
1132 To employ /proc/pid/pagemap efficiently, use /proc/pid/maps to
1134 determine which areas of memory are actually mapped and seek to
1135 skip over unmapped regions.
1136 The /proc/pid/pagemap file is present only if the CON‐
1138 FIG_PROC_PAGE_MONITOR kernel configuration option is enabled.
1139 Permission to access this file is governed by a ptrace access
1141 mode PTRACE_MODE_READ_FSCREDS check; see ptrace(2).
1142 /proc/pid/personality (since Linux 2.6.28)
1144 This read-only file exposes the process's execution domain, as
1145 set by personality(2). The value is displayed in hexadecimal
1146 notation.
1147 Permission to access this file is governed by a ptrace access
1149 mode PTRACE_MODE_ATTACH_FSCREDS check; see ptrace(2).
1150 /proc/pid/root
1152 UNIX and Linux support the idea of a per-process root of the
1153 filesystem, set by the chroot(2) system call. This file is a
1154 symbolic link that points to the process's root directory, and
1155 behaves in the same way as exe, and fd/*.
1156 Note however that this file is not merely a symbolic link. It
1158 provides the same view of the filesystem (including namespaces
1159 and the set of per-process mounts) as the process itself. An
1160 example illustrates this point. In one terminal, we start a
1161 shell in new user and mount namespaces, and in that shell we
1162 create some new mounts:
1163 $ PS1='sh1# ' unshare -Urnm
1165 sh1# mount -t tmpfs tmpfs /etc # Mount empty tmpfs at /etc
1166 sh1# mount --bind /usr /dev # Mount /usr at /dev
1167 sh1# echo $$
1168 27123
1169 In a second terminal window, in the initial mount namespace, we
1171 look at the contents of the corresponding mounts in the initial
1172 and new namespaces:
1173 $ PS1='sh2# ' sudo sh
1175 sh2# ls /etc | wc -l # In initial NS
1176 309
1177 sh2# ls /proc/27123/root/etc | wc -l # /etc in other NS
1178 0 # The empty tmpfs dir
1179 sh2# ls /dev | wc -l # In initial NS
1180 205
1181 sh2# ls /proc/27123/root/dev | wc -l # /dev in other NS
1182 11 # Actually bind
1183 # mounted to /usr
1184 sh2# ls /usr | wc -l # /usr in initial NS
1185 11
1186 In a multithreaded process, the contents of the /proc/pid/root
1188 symbolic link are not available if the main thread has already
1189 terminated (typically by calling pthread_exit(3)).
1190 Permission to dereference or read (readlink(2)) this symbolic
1192 link is governed by a ptrace access mode PTRACE_MODE_READ_FS‐
1193 CREDS check; see ptrace(2).
1194 /proc/pid/projid_map (since Linux 3.7)
1196 See user_namespaces(7).
1197 /proc/pid/seccomp (Linux 2.6.12 to Linux 2.6.22)
1199 This file can be used to read and change the process's secure
1200 computing (seccomp) mode setting. It contains the value 0 if
1201 the process is not in seccomp mode, and 1 if the process is in
1202 strict seccomp mode (see seccomp(2)). Writing 1 to this file
1203 places the process irreversibly in strict seccomp mode. (Fur‐
1204 ther attempts to write to the file fail with the EPERM error.)
1205 In Linux 2.6.23, this file went away, to be replaced by the
1207 prctl(2) PR_GET_SECCOMP and PR_SET_SECCOMP operations (and later
1208 by seccomp(2) and the Seccomp field in /proc/pid/status).
1209 /proc/pid/setgroups (since Linux 3.19)
1211 See user_namespaces(7).
1212 /proc/pid/smaps (since Linux 2.6.14)
1214 This file shows memory consumption for each of the process's
1215 mappings. (The pmap(1) command displays similar information, in
1216 a form that may be easier for parsing.) For each mapping there
1217 is a series of lines such as the following:
1218 00400000-0048a000 r-xp 00000000 fd:03 960637 /bin/bash
1220 Size: 552 kB
1221 Rss: 460 kB
1222 Pss: 100 kB
1223 Shared_Clean: 452 kB
1224 Shared_Dirty: 0 kB
1225 Private_Clean: 8 kB
1226 Private_Dirty: 0 kB
1227 Referenced: 460 kB
1228 Anonymous: 0 kB
1229 AnonHugePages: 0 kB
1230 ShmemHugePages: 0 kB
1231 ShmemPmdMapped: 0 kB
1232 Swap: 0 kB
1233 KernelPageSize: 4 kB
1234 MMUPageSize: 4 kB
1235 Locked: 0 kB
1236 ProtectionKey: 0
1237 VmFlags: rd ex mr mw me dw
1238 The first of these lines shows the same information as is dis‐
1240 played for the mapping in /proc/pid/maps. The following lines
1241 show the size of the mapping, the amount of the mapping that is
1242 currently resident in RAM ("Rss"), the process's proportional
1243 share of this mapping ("Pss"), the number of clean and dirty
1244 shared pages in the mapping, and the number of clean and dirty
1245 private pages in the mapping. "Referenced" indicates the amount
1246 of memory currently marked as referenced or accessed. "Anony‐
1247 mous" shows the amount of memory that does not belong to any
1248 file. "Swap" shows how much would-be-anonymous memory is also
1249 used, but out on swap.
1250 The "KernelPageSize" line (available since Linux 2.6.29) is the
1252 page size used by the kernel to back the virtual memory area.
1253 This matches the size used by the MMU in the majority of cases.
1254 However, one counter-example occurs on PPC64 kernels whereby a
1255 kernel using 64 kB as a base page size may still use 4 kB pages
1256 for the MMU on older processors. To distinguish the two at‐
1257 tributes, the "MMUPageSize" line (also available since Linux
1258 2.6.29) reports the page size used by the MMU.
1259 The "Locked" indicates whether the mapping is locked in memory
1261 or not.
1262 The "ProtectionKey" line (available since Linux 4.9, on x86
1264 only) contains the memory protection key (see pkeys(7)) associ‐
1265 ated with the virtual memory area. This entry is present only
1266 if the kernel was built with the CONFIG_X86_INTEL_MEMORY_PROTEC‐
1267 TION_KEYS configuration option (since Linux 4.6).
1268 The "VmFlags" line (available since Linux 3.8) represents the
1270 kernel flags associated with the virtual memory area, encoded
1271 using the following two-letter codes:
1272 rd - readable
1274 wr - writable
1275 ex - executable
1276 sh - shared
1277 mr - may read
1278 mw - may write
1279 me - may execute
1280 ms - may share
1281 gd - stack segment grows down
1282 pf - pure PFN range
1283 dw - disabled write to the mapped file
1284 lo - pages are locked in memory
1285 io - memory mapped I/O area
1286 sr - sequential read advise provided
1287 rr - random read advise provided
1288 dc - do not copy area on fork
1289 de - do not expand area on remapping
1290 ac - area is accountable
1291 nr - swap space is not reserved for the area
1292 ht - area uses huge tlb pages
1293 sf - perform synchronous page faults (since Linux 4.15)
1294 nl - non-linear mapping (removed in Linux 4.0)
1295 ar - architecture specific flag
1296 wf - wipe on fork (since Linux 4.14)
1297 dd - do not include area into core dump
1298 sd - soft-dirty flag (since Linux 3.13)
1299 mm - mixed map area
1300 hg - huge page advise flag
1301 nh - no-huge page advise flag
1302 mg - mergeable advise flag
1303 um - userfaultfd missing pages tracking (since Linux 4.3)
1304 uw - userfaultfd wprotect pages tracking (since Linux 4.3)
1305 The /proc/pid/smaps file is present only if the CON‐
1307 FIG_PROC_PAGE_MONITOR kernel configuration option is enabled.
1308 /proc/pid/stack (since Linux 2.6.29)
1310 This file provides a symbolic trace of the function calls in
1311 this process's kernel stack. This file is provided only if the
1312 kernel was built with the CONFIG_STACKTRACE configuration op‐
1313 tion.
1314 Permission to access this file is governed by a ptrace access
1316 mode PTRACE_MODE_ATTACH_FSCREDS check; see ptrace(2).
1317 /proc/pid/stat
1319 Status information about the process. This is used by ps(1).
1320 It is defined in the kernel source file fs/proc/array.c.
1321 The fields, in order, with their proper scanf(3) format speci‐
1323 fiers, are listed below. Whether or not certain of these fields
1324 display valid information is governed by a ptrace access mode
1325 PTRACE_MODE_READ_FSCREDS | PTRACE_MODE_NOAUDIT check (refer to
1326 ptrace(2)). If the check denies access, then the field value is
1327 displayed as 0. The affected fields are indicated with the
1328 marking [PT].
1329 (1) pid %d
1331 The process ID.
1332 (2) comm %s
1334 The filename of the executable, in parentheses. Strings
1335 longer than TASK_COMM_LEN (16) characters (including the
1336 terminating null byte) are silently truncated. This is
1337 visible whether or not the executable is swapped out.
1338 (3) state %c
1340 One of the following characters, indicating process
1341 state:
1342 R Running
1344 S Sleeping in an interruptible wait
1346 D Waiting in uninterruptible disk sleep
1348 Z Zombie
1350 T Stopped (on a signal) or (before Linux 2.6.33)
1352 trace stopped
1353 t Tracing stop (Linux 2.6.33 onward)
1355 W Paging (only before Linux 2.6.0)
1357 X Dead (from Linux 2.6.0 onward)
1359 x Dead (Linux 2.6.33 to 3.13 only)
1361 K Wakekill (Linux 2.6.33 to 3.13 only)
1363 W Waking (Linux 2.6.33 to 3.13 only)
1365 P Parked (Linux 3.9 to 3.13 only)
1367 I Idle (Linux 4.14 onward)
1369 (4) ppid %d
1371 The PID of the parent of this process.
1372 (5) pgrp %d
1374 The process group ID of the process.
1375 (6) session %d
1377 The session ID of the process.
1378 (7) tty_nr %d
1380 The controlling terminal of the process. (The minor de‐
1381 vice number is contained in the combination of bits 31 to
1382 20 and 7 to 0; the major device number is in bits 15 to
1383 8.)
1384 (8) tpgid %d
1386 The ID of the foreground process group of the controlling
1387 terminal of the process.
1388 (9) flags %u
1390 The kernel flags word of the process. For bit meanings,
1391 see the PF_* defines in the Linux kernel source file in‐
1392 clude/linux/sched.h. Details depend on the kernel ver‐
1393 sion.
1394 The format for this field was %lu before Linux 2.6.
1396 (10) minflt %lu
1398 The number of minor faults the process has made which
1399 have not required loading a memory page from disk.
1400 (11) cminflt %lu
1402 The number of minor faults that the process's waited-for
1403 children have made.
1404 (12) majflt %lu
1406 The number of major faults the process has made which
1407 have required loading a memory page from disk.
1408 (13) cmajflt %lu
1410 The number of major faults that the process's waited-for
1411 children have made.
1412 (14) utime %lu
1414 Amount of time that this process has been scheduled in
1415 user mode, measured in clock ticks (divide by
1416 sysconf(_SC_CLK_TCK)). This includes guest time,
1417 guest_time (time spent running a virtual CPU, see below),
1418 so that applications that are not aware of the guest time
1419 field do not lose that time from their calculations.
1420 (15) stime %lu
1422 Amount of time that this process has been scheduled in
1423 kernel mode, measured in clock ticks (divide by
1424 sysconf(_SC_CLK_TCK)).
1425 (16) cutime %ld
1427 Amount of time that this process's waited-for children
1428 have been scheduled in user mode, measured in clock ticks
1429 (divide by sysconf(_SC_CLK_TCK)). (See also times(2).)
1430 This includes guest time, cguest_time (time spent running
1431 a virtual CPU, see below).
1432 (17) cstime %ld
1434 Amount of time that this process's waited-for children
1435 have been scheduled in kernel mode, measured in clock
1436 ticks (divide by sysconf(_SC_CLK_TCK)).
1437 (18) priority %ld
1439 (Explanation for Linux 2.6) For processes running a real-
1440 time scheduling policy (policy below; see sched_setsched‐
1441 uler(2)), this is the negated scheduling priority, minus
1442 one; that is, a number in the range -2 to -100, corre‐
1443 sponding to real-time priorities 1 to 99. For processes
1444 running under a non-real-time scheduling policy, this is
1445 the raw nice value (setpriority(2)) as represented in the
1446 kernel. The kernel stores nice values as numbers in the
1447 range 0 (high) to 39 (low), corresponding to the user-
1448 visible nice range of -20 to 19.
1449 Before Linux 2.6, this was a scaled value based on the
1451 scheduler weighting given to this process.
1452 (19) nice %ld
1454 The nice value (see setpriority(2)), a value in the range
1455 19 (low priority) to -20 (high priority).
1456 (20) num_threads %ld
1458 Number of threads in this process (since Linux 2.6). Be‐
1459 fore Linux 2.6, this field was hard coded to 0 as a
1460 placeholder for an earlier removed field.
1461 (21) itrealvalue %ld
1463 The time in jiffies before the next SIGALRM is sent to
1464 the process due to an interval timer. Since Linux
1465 2.6.17, this field is no longer maintained, and is hard
1466 coded as 0.
1467 (22) starttime %llu
1469 The time the process started after system boot. Before
1470 Linux 2.6, this value was expressed in jiffies. Since
1471 Linux 2.6, the value is expressed in clock ticks (divide
1472 by sysconf(_SC_CLK_TCK)).
1473 The format for this field was %lu before Linux 2.6.
1475 (23) vsize %lu
1477 Virtual memory size in bytes.
1478 (24) rss %ld
1480 Resident Set Size: number of pages the process has in
1481 real memory. This is just the pages which count toward
1482 text, data, or stack space. This does not include pages
1483 which have not been demand-loaded in, or which are
1484 swapped out. This value is inaccurate; see
1485 /proc/pid/statm below.
1486 (25) rsslim %lu
1488 Current soft limit in bytes on the rss of the process;
1489 see the description of RLIMIT_RSS in getrlimit(2).
1490 (26) startcode %lu [PT]
1492 The address above which program text can run.
1493 (27) endcode %lu [PT]
1495 The address below which program text can run.
1496 (28) startstack %lu [PT]
1498 The address of the start (i.e., bottom) of the stack.
1499 (29) kstkesp %lu [PT]
1501 The current value of ESP (stack pointer), as found in the
1502 kernel stack page for the process.
1503 (30) kstkeip %lu [PT]
1505 The current EIP (instruction pointer).
1506 (31) signal %lu
1508 The bitmap of pending signals, displayed as a decimal
1509 number. Obsolete, because it does not provide informa‐
1510 tion on real-time signals; use /proc/pid/status instead.
1511 (32) blocked %lu
1513 The bitmap of blocked signals, displayed as a decimal
1514 number. Obsolete, because it does not provide informa‐
1515 tion on real-time signals; use /proc/pid/status instead.
1516 (33) sigignore %lu
1518 The bitmap of ignored signals, displayed as a decimal
1519 number. Obsolete, because it does not provide informa‐
1520 tion on real-time signals; use /proc/pid/status instead.
1521 (34) sigcatch %lu
1523 The bitmap of caught signals, displayed as a decimal num‐
1524 ber. Obsolete, because it does not provide information
1525 on real-time signals; use /proc/pid/status instead.
1526 (35) wchan %lu [PT]
1528 This is the "channel" in which the process is waiting.
1529 It is the address of a location in the kernel where the
1530 process is sleeping. The corresponding symbolic name can
1531 be found in /proc/pid/wchan.
1532 (36) nswap %lu
1534 Number of pages swapped (not maintained).
1535 (37) cnswap %lu
1537 Cumulative nswap for child processes (not maintained).
1538 (38) exit_signal %d (since Linux 2.1.22)
1540 Signal to be sent to parent when we die.
1541 (39) processor %d (since Linux 2.2.8)
1543 CPU number last executed on.
1544 (40) rt_priority %u (since Linux 2.5.19)
1546 Real-time scheduling priority, a number in the range 1 to
1547 99 for processes scheduled under a real-time policy, or
1548 0, for non-real-time processes (see sched_setsched‐
1549 uler(2)).
1550 (41) policy %u (since Linux 2.5.19)
1552 Scheduling policy (see sched_setscheduler(2)). Decode
1553 using the SCHED_* constants in linux/sched.h.
1554 The format for this field was %lu before Linux 2.6.22.
1556 (42) delayacct_blkio_ticks %llu (since Linux 2.6.18)
1558 Aggregated block I/O delays, measured in clock ticks
1559 (centiseconds).
1560 (43) guest_time %lu (since Linux 2.6.24)
1562 Guest time of the process (time spent running a virtual
1563 CPU for a guest operating system), measured in clock
1564 ticks (divide by sysconf(_SC_CLK_TCK)).
1565 (44) cguest_time %ld (since Linux 2.6.24)
1567 Guest time of the process's children, measured in clock
1568 ticks (divide by sysconf(_SC_CLK_TCK)).
1569 (45) start_data %lu (since Linux 3.3) [PT]
1571 Address above which program initialized and uninitialized
1572 (BSS) data are placed.
1573 (46) end_data %lu (since Linux 3.3) [PT]
1575 Address below which program initialized and uninitialized
1576 (BSS) data are placed.
1577 (47) start_brk %lu (since Linux 3.3) [PT]
1579 Address above which program heap can be expanded with
1580 brk(2).
1581 (48) arg_start %lu (since Linux 3.5) [PT]
1583 Address above which program command-line arguments (argv)
1584 are placed.
1585 (49) arg_end %lu (since Linux 3.5) [PT]
1587 Address below program command-line arguments (argv) are
1588 placed.
1589 (50) env_start %lu (since Linux 3.5) [PT]
1591 Address above which program environment is placed.
1592 (51) env_end %lu (since Linux 3.5) [PT]
1594 Address below which program environment is placed.
1595 (52) exit_code %d (since Linux 3.5) [PT]
1597 The thread's exit status in the form reported by wait‐
1598 pid(2).
1599 /proc/pid/statm
1601 Provides information about memory usage, measured in pages. The
1602 columns are:
1603 size (1) total program size
1605 (same as VmSize in /proc/pid/status)
1606 resident (2) resident set size
1607 (inaccurate; same as VmRSS in /proc/pid/status)
1608 shared (3) number of resident shared pages
1609 (i.e., backed by a file)
1610 (inaccurate; same as RssFile+RssShmem in
1611 /proc/pid/status)
1612 text (4) text (code)
1613 lib (5) library (unused since Linux 2.6; always 0)
1614 data (6) data + stack
1615 dt (7) dirty pages (unused since Linux 2.6; always 0)
1616 Some of these values are inaccurate because of a kernel-internal
1618 scalability optimization. If accurate values are required, use
1619 /proc/pid/smaps or /proc/pid/smaps_rollup instead, which are
1620 much slower but provide accurate, detailed information.
1621 /proc/pid/status
1623 Provides much of the information in /proc/pid/stat and
1624 /proc/pid/statm in a format that's easier for humans to parse.
1625 Here's an example:
1626 $ cat /proc/$$/status
1628 Name: bash
1629 Umask: 0022
1630 State: S (sleeping)
1631 Tgid: 17248
1632 Ngid: 0
1633 Pid: 17248
1634 PPid: 17200
1635 TracerPid: 0
1636 Uid: 1000 1000 1000 1000
1637 Gid: 100 100 100 100
1638 FDSize: 256
1639 Groups: 16 33 100
1640 NStgid: 17248
1641 NSpid: 17248
1642 NSpgid: 17248
1643 NSsid: 17200
1644 VmPeak: 131168 kB
1645 VmSize: 131168 kB
1646 VmLck: 0 kB
1647 VmPin: 0 kB
1648 VmHWM: 13484 kB
1649 VmRSS: 13484 kB
1650 RssAnon: 10264 kB
1651 RssFile: 3220 kB
1652 RssShmem: 0 kB
1653 VmData: 10332 kB
1654 VmStk: 136 kB
1655 VmExe: 992 kB
1656 VmLib: 2104 kB
1657 VmPTE: 76 kB
1658 VmPMD: 12 kB
1659 VmSwap: 0 kB
1660 HugetlbPages: 0 kB # 4.4
1661 CoreDumping: 0 # 4.15
1662 Threads: 1
1663 SigQ: 0/3067
1664 SigPnd: 0000000000000000
1665 ShdPnd: 0000000000000000
1666 SigBlk: 0000000000010000
1667 SigIgn: 0000000000384004
1668 SigCgt: 000000004b813efb
1669 CapInh: 0000000000000000
1670 CapPrm: 0000000000000000
1671 CapEff: 0000000000000000
1672 CapBnd: ffffffffffffffff
1673 CapAmb: 0000000000000000
1674 NoNewPrivs: 0
1675 Seccomp: 0
1676 Speculation_Store_Bypass: vulnerable
1677 Cpus_allowed: 00000001
1678 Cpus_allowed_list: 0
1679 Mems_allowed: 1
1680 Mems_allowed_list: 0
1681 voluntary_ctxt_switches: 150
1682 nonvoluntary_ctxt_switches: 545
1683 The fields are as follows:
1685 Name Command run by this process. Strings longer than
1687 TASK_COMM_LEN [22m(16) characters (including the terminating
1688 null byte) are silently truncated.
1689 Umask Process umask, expressed in octal with a leading zero;
1691 see umask(2). (Since Linux 4.7.)
1692 State Current state of the process. One of "R (running)", "S
1694 (sleeping)", "D (disk sleep)", "T (stopped)", "t (tracing
1695 stop)", "Z (zombie)", or "X (dead)".
1696 Tgid Thread group ID (i.e., Process ID).
1698 Ngid NUMA group ID (0 if none; since Linux 3.13).
1700 Pid Thread ID (see gettid(2)).
1702 PPid PID of parent process.
1704 TracerPid
1706 PID of process tracing this process (0 if not being
1707 traced).
1708 Uid, Gid
1710 Real, effective, saved set, and filesystem UIDs (GIDs).
1711 FDSize Number of file descriptor slots currently allocated.
1713 Groups Supplementary group list.
1715 NStgid Thread group ID (i.e., PID) in each of the PID namespaces
1717 of which pid is a member. The leftmost entry shows the
1718 value with respect to the PID namespace of the process
1719 that mounted this procfs (or the root namespace if
1720 mounted by the kernel), followed by the value in succes‐
1721 sively nested inner namespaces. (Since Linux 4.1.)
1722 NSpid Thread ID in each of the PID namespaces of which pid is a
1724 member. The fields are ordered as for NStgid. (Since
1725 Linux 4.1.)
1726 NSpgid Process group ID in each of the PID namespaces of which
1728 pid is a member. The fields are ordered as for NStgid.
1729 (Since Linux 4.1.)
1730 NSsid descendant namespace session ID hierarchy Session ID in
1732 each of the PID namespaces of which pid is a member. The
1733 fields are ordered as for NStgid. (Since Linux 4.1.)
1734 VmPeak Peak virtual memory size.
1736 VmSize Virtual memory size.
1738 VmLck Locked memory size (see mlock(2)).
1740 VmPin Pinned memory size (since Linux 3.2). These are pages
1742 that can't be moved because something needs to directly
1743 access physical memory.
1744 VmHWM Peak resident set size ("high water mark"). This value
1746 is inaccurate; see /proc/pid/statm above.
1747 VmRSS Resident set size. Note that the value here is the sum
1749 of RssAnon, RssFile, and RssShmem. This value is inaccu‐
1750 rate; see /proc/pid/statm above.
1751 RssAnon
1753 Size of resident anonymous memory. (since Linux 4.5).
1754 This value is inaccurate; see /proc/pid/statm above.
1755 RssFile
1757 Size of resident file mappings. (since Linux 4.5). This
1758 value is inaccurate; see /proc/pid/statm above.
1759 RssShmem
1761 Size of resident shared memory (includes System V shared
1762 memory, mappings from tmpfs(5), and shared anonymous map‐
1763 pings). (since Linux 4.5).
1764 VmData, VmStk, VmExe
1766 Size of data, stack, and text segments. This value is
1767 inaccurate; see /proc/pid/statm above.
1768 VmLib Shared library code size.
1770 VmPTE Page table entries size (since Linux 2.6.10).
1772 VmPMD Size of second-level page tables (added in Linux 4.0; re‐
1774 moved in Linux 4.15).
1775 VmSwap Swapped-out virtual memory size by anonymous private
1777 pages; shmem swap usage is not included (since Linux
1778 2.6.34). This value is inaccurate; see /proc/pid/statm
1779 above.
1780 HugetlbPages
1782 Size of hugetlb memory portions (since Linux 4.4).
1783 CoreDumping
1785 Contains the value 1 if the process is currently dumping
1786 core, and 0 if it is not (since Linux 4.15). This infor‐
1787 mation can be used by a monitoring process to avoid
1788 killing a process that is currently dumping core, which
1789 could result in a corrupted core dump file.
1790 Threads
1792 Number of threads in process containing this thread.
1793 SigQ This field contains two slash-separated numbers that re‐
1795 late to queued signals for the real user ID of this
1796 process. The first of these is the number of currently
1797 queued signals for this real user ID, and the second is
1798 the resource limit on the number of queued signals for
1799 this process (see the description of RLIMIT_SIGPENDING in
1800 getrlimit(2)).
1801 SigPnd, ShdPnd
1803 Mask (expressed in hexadecimal) of signals pending for
1804 thread and for process as a whole (see pthreads(7) and
1805 signal(7)).
1806 SigBlk, SigIgn, SigCgt
1808 Masks (expressed in hexadecimal) indicating signals being
1809 blocked, ignored, and caught (see signal(7)).
1810 CapInh, CapPrm, CapEff
1812 Masks (expressed in hexadecimal) of capabilities enabled
1813 in inheritable, permitted, and effective sets (see capa‐
1814 bilities(7)).
1815 CapBnd Capability bounding set, expressed in hexadecimal (since
1817 Linux 2.6.26, see capabilities(7)).
1818 CapAmb Ambient capability set, expressed in hexadecimal (since
1820 Linux 4.3, see capabilities(7)).
1821 NoNewPrivs
1823 Value of the no_new_privs bit (since Linux 4.10, see
1824 prctl(2)).
1825 Seccomp
1827 Seccomp mode of the process (since Linux 3.8, see sec‐
1828 comp(2)). 0 means SECCOMP_MODE_DISABLED; 1 means SEC‐
1829 COMP_MODE_STRICT; 2 means SECCOMP_MODE_FILTER. This
1830 field is provided only if the kernel was built with the
1831 CONFIG_SECCOMP kernel configuration option enabled.
1832 Speculation_Store_Bypass
1834 Speculation flaw mitigation state (since Linux 4.17, see
1835 prctl(2)).
1836 Cpus_allowed
1838 Hexadecimal mask of CPUs on which this process may run
1839 (since Linux 2.6.24, see cpuset(7)).
1840 Cpus_allowed_list
1842 Same as previous, but in "list format" (since Linux
1843 2.6.26, see cpuset(7)).
1844 Mems_allowed
1846 Mask of memory nodes allowed to this process (since Linux
1847 2.6.24, see cpuset(7)).
1848 Mems_allowed_list
1850 Same as previous, but in "list format" (since Linux
1851 2.6.26, see cpuset(7)).
1852 voluntary_ctxt_switches, nonvoluntary_ctxt_switches
1854 Number of voluntary and involuntary context switches
1855 (since Linux 2.6.23).
1856 /proc/pid/syscall (since Linux 2.6.27)
1858 This file exposes the system call number and argument registers
1859 for the system call currently being executed by the process,
1860 followed by the values of the stack pointer and program counter
1861 registers. The values of all six argument registers are ex‐
1862 posed, although most system calls use fewer registers.
1863 If the process is blocked, but not in a system call, then the
1865 file displays -1 in place of the system call number, followed by
1866 just the values of the stack pointer and program counter. If
1867 process is not blocked, then the file contains just the string
1868 "running".
1869 This file is present only if the kernel was configured with CON‐
1871 FIG_HAVE_ARCH_TRACEHOOK.
1872 Permission to access this file is governed by a ptrace access
1874 mode PTRACE_MODE_ATTACH_FSCREDS check; see ptrace(2).
1875 /proc/pid/task (since Linux 2.6.0)
1877 This is a directory that contains one subdirectory for each
1878 thread in the process. The name of each subdirectory is the nu‐
1879 merical thread ID (tid) of the thread (see gettid(2)).
1880 Within each of these subdirectories, there is a set of files
1882 with the same names and contents as under the /proc/pid directo‐
1883 ries. For attributes that are shared by all threads, the con‐
1884 tents for each of the files under the task/tid subdirectories
1885 will be the same as in the corresponding file in the parent
1886 /proc/pid directory (e.g., in a multithreaded process, all of
1887 the task/tid/cwd files will have the same value as the
1888 /proc/pid/cwd file in the parent directory, since all of the
1889 threads in a process share a working directory). For attributes
1890 that are distinct for each thread, the corresponding files under
1891 task/tid may have different values (e.g., various fields in each
1892 of the task/tid/status files may be different for each thread),
1893 or they might not exist in /proc/pid at all.
1894 In a multithreaded process, the contents of the /proc/pid/task
1896 directory are not available if the main thread has already ter‐
1897 minated (typically by calling pthread_exit(3)).
1898 /proc/pid/task/tid/children (since Linux 3.5)
1900 A space-separated list of child tasks of this task. Each child
1901 task is represented by its TID.
1902 This option is intended for use by the checkpoint-restore (CRIU)
1904 system, and reliably provides a list of children only if all of
1905 the child processes are stopped or frozen. It does not work
1906 properly if children of the target task exit while the file is
1907 being read! Exiting children may cause non-exiting children to
1908 be omitted from the list. This makes this interface even more
1909 unreliable than classic PID-based approaches if the inspected
1910 task and its children aren't frozen, and most code should proba‐
1911 bly not use this interface.
1912 Until Linux 4.2, the presence of this file was governed by the
1914 CONFIG_CHECKPOINT_RESTORE kernel configuration option. Since
1915 Linux 4.2, it is governed by the CONFIG_PROC_CHILDREN option.
1916 /proc/pid/timers (since Linux 3.10)
1918 A list of the POSIX timers for this process. Each timer is
1919 listed with a line that starts with the string "ID:". For exam‐
1920 ple:
1921 ID: 1
1923 signal: 60/00007fff86e452a8
1924 notify: signal/pid.2634
1925 ClockID: 0
1926 ID: 0
1927 signal: 60/00007fff86e452a8
1928 notify: signal/pid.2634
1929 ClockID: 1
1930 The lines shown for each timer have the following meanings:
1932 ID The ID for this timer. This is not the same as the timer
1934 ID returned by timer_create(2); rather, it is the same
1935 kernel-internal ID that is available via the si_timerid
1936 field of the siginfo_t structure (see sigaction(2)).
1937 signal This is the signal number that this timer uses to deliver
1939 notifications followed by a slash, and then the
1940 sigev_value value supplied to the signal handler. Valid
1941 only for timers that notify via a signal.
1942 notify The part before the slash specifies the mechanism that
1944 this timer uses to deliver notifications, and is one of
1945 "thread", "signal", or "none". Immediately following the
1946 slash is either the string "tid" for timers with
1947 SIGEV_THREAD_ID notification, or "pid" for timers that
1948 notify by other mechanisms. Following the "." is the PID
1949 of the process (or the kernel thread ID of the thread)
1950 that will be delivered a signal if the timer delivers no‐
1951 tifications via a signal.
1952 ClockID
1954 This field identifies the clock that the timer uses for
1955 measuring time. For most clocks, this is a number that
1956 matches one of the user-space CLOCK_* constants exposed
1957 via <time.h>. CLOCK_PROCESS_CPUTIME_ID timers display
1958 with a value of -6 in this field.
1959 CLOCK_THREAD_CPUTIME_ID timers display with a value of -2
1960 in this field.
1961 This file is available only when the kernel was configured with
1963 CONFIG_CHECKPOINT_RESTORE.
1964 /proc/pid/timerslack_ns (since Linux 4.6)
1966 This file exposes the process's "current" timer slack value, ex‐
1967 pressed in nanoseconds. The file is writable, allowing the
1968 process's timer slack value to be changed. Writing 0 to this
1969 file resets the "current" timer slack to the "default" timer
1970 slack value. For further details, see the discussion of
1971 PR_SET_TIMERSLACK in prctl(2).
1972 Initially, permission to access this file was governed by a
1974 ptrace access mode PTRACE_MODE_ATTACH_FSCREDS check (see
1975 ptrace(2)). However, this was subsequently deemed too strict a
1976 requirement (and had the side effect that requiring a process to
1977 have the CAP_SYS_PTRACE capability would also allow it to view
1978 and change any process's memory). Therefore, since Linux 4.9,
1979 only the (weaker) CAP_SYS_NICE capability is required to access
1980 this file.
1981 /proc/pid/uid_map (since Linux 3.5)
1983 See user_namespaces(7).
1984 /proc/pid/wchan (since Linux 2.6.0)
1986 The symbolic name corresponding to the location in the kernel
1987 where the process is sleeping.
1988 Permission to access this file is governed by a ptrace access
1990 mode PTRACE_MODE_READ_FSCREDS check; see ptrace(2).
1991 /proc/tid
1993 There is a numerical subdirectory for each running thread that
1994 is not a thread group leader (i.e., a thread whose thread ID is
1995 not the same as its process ID); the subdirectory is named by
1996 the thread ID. Each one of these subdirectories contains files
1997 and subdirectories exposing information about the thread with
1998 the thread ID tid. The contents of these directories are the
1999 same as the corresponding /proc/pid/task/tid directories.
2000 The /proc/tid subdirectories are not visible when iterating
2002 through /proc with getdents(2) (and thus are not visible when
2003 one uses ls(1) to view the contents of /proc). However, the
2004 pathnames of these directories are visible to (i.e., usable as
2005 arguments in) system calls that operate on pathnames.
2006 /proc/apm
2008 Advanced power management version and battery information when
2009 CONFIG_APM is defined at kernel compilation time.
2010 /proc/buddyinfo
2012 This file contains information which is used for diagnosing mem‐
2013 ory fragmentation issues. Each line starts with the identifica‐
2014 tion of the node and the name of the zone which together iden‐
2015 tify a memory region. This is then followed by the count of
2016 available chunks of a certain order in which these zones are
2017 split. The size in bytes of a certain order is given by the
2018 formula:
2019 (2^order) * PAGE_SIZE
2021 The binary buddy allocator algorithm inside the kernel will
2023 split one chunk into two chunks of a smaller order (thus with
2024 half the size) or combine two contiguous chunks into one larger
2025 chunk of a higher order (thus with double the size) to satisfy
2026 allocation requests and to counter memory fragmentation. The
2027 order matches the column number, when starting to count at zero.
2028 For example on an x86-64 system:
2030 Node 0, zone DMA 1 1 1 0 2 1 1 0 1 1 3
2031 Node 0, zone DMA32 65 47 4 81 52 28 13 10 5 1 404
2032 Node 0, zone Normal 216 55 189 101 84 38 37 27 5 3 587
2033 In this example, there is one node containing three zones and
2035 there are 11 different chunk sizes. If the page size is 4 kilo‐
2036 bytes, then the first zone called DMA (on x86 the first 16
2037 megabyte of memory) has 1 chunk of 4 kilobytes (order 0) avail‐
2038 able and has 3 chunks of 4 megabytes (order 10) available.
2039 If the memory is heavily fragmented, the counters for higher or‐
2041 der chunks will be zero and allocation of large contiguous areas
2042 will fail.
2043 Further information about the zones can be found in /proc/zone‐
2045 info.
2046 /proc/bus
2048 Contains subdirectories for installed buses.
2049 /proc/bus/pccard
2051 Subdirectory for PCMCIA devices when CONFIG_PCMCIA is set at
2052 kernel compilation time.
2053 /proc/bus/pccard/drivers
2055 /proc/bus/pci
2057 Contains various bus subdirectories and pseudo-files containing
2058 information about PCI buses, installed devices, and device driv‐
2059 ers. Some of these files are not ASCII.
2060 /proc/bus/pci/devices
2062 Information about PCI devices. They may be accessed through
2063 lspci(8) and setpci(8).
2064 /proc/cgroups (since Linux 2.6.24)
2066 See cgroups(7).
2067 /proc/cmdline
2069 Arguments passed to the Linux kernel at boot time. Often done
2070 via a boot manager such as lilo(8) or grub(8).
2071 /proc/config.gz (since Linux 2.6)
2073 This file exposes the configuration options that were used to
2074 build the currently running kernel, in the same format as they
2075 would be shown in the .config file that resulted when configur‐
2076 ing the kernel (using make xconfig, make config, or similar).
2077 The file contents are compressed; view or search them using
2078 zcat(1) and zgrep(1). As long as no changes have been made to
2079 the following file, the contents of /proc/config.gz are the same
2080 as those provided by:
2081 cat /lib/modules/$(uname -r)/build/.config
2083 /proc/config.gz is provided only if the kernel is configured
2085 with CONFIG_IKCONFIG_PROC.
2086 /proc/crypto
2088 A list of the ciphers provided by the kernel crypto API. For
2089 details, see the kernel Linux Kernel Crypto API documentation
2090 available under the kernel source directory Documenta‐
2091 tion/crypto/ (or Documentation/DocBook before Linux 4.10; the
2092 documentation can be built using a command such as make htmldocs
2093 in the root directory of the kernel source tree).
2094 /proc/cpuinfo
2096 This is a collection of CPU and system architecture dependent
2097 items, for each supported architecture a different list. Two
2098 common entries are processor which gives CPU number and bo‐
2099 gomips; a system constant that is calculated during kernel ini‐
2100 tialization. SMP machines have information for each CPU. The
2101 lscpu(1) command gathers its information from this file.
2102 /proc/devices
2104 Text listing of major numbers and device groups. This can be
2105 used by MAKEDEV scripts for consistency with the kernel.
2106 /proc/diskstats (since Linux 2.5.69)
2108 This file contains disk I/O statistics for each disk device.
2109 See the Linux kernel source file Documentation/ad‐
2110 min-guide/iostats.rst (or Documentation/iostats.txt before Linux
2111 5.3) for further information.
2112 /proc/dma
2114 This is a list of the registered ISA DMA (direct memory access)
2115 channels in use.
2116 /proc/driver
2118 Empty subdirectory.
2119 /proc/execdomains
2121 List of the execution domains (ABI personalities).
2122 /proc/fb
2124 Frame buffer information when CONFIG_FB is defined during kernel
2125 compilation.
2126 /proc/filesystems
2128 A text listing of the filesystems which are supported by the
2129 kernel, namely filesystems which were compiled into the kernel
2130 or whose kernel modules are currently loaded. (See also
2131 filesystems(5).) If a filesystem is marked with "nodev", this
2132 means that it does not require a block device to be mounted
2133 (e.g., virtual filesystem, network filesystem).
2134 Incidentally, this file may be used by mount(8) when no filesys‐
2136 tem is specified and it didn't manage to determine the filesys‐
2137 tem type. Then filesystems contained in this file are tried
2138 (excepted those that are marked with "nodev").
2139 /proc/fs
2141 Contains subdirectories that in turn contain files with informa‐
2142 tion about (certain) mounted filesystems.
2143 /proc/ide
2145 This directory exists on systems with the IDE bus. There are
2146 directories for each IDE channel and attached device. Files in‐
2147 clude:
2148 cache buffer size in KB
2150 capacity number of sectors
2151 driver driver version
2152 geometry physical and logical geometry
2153 identify in hexadecimal
2154 media media type
2155 model manufacturer's model number
2156 settings drive settings
2157 smart_thresholds IDE disk management thresholds (in hex)
2158 smart_values IDE disk management values (in hex)
2159 The hdparm(8) utility provides access to this information in a
2161 friendly format.
2162 /proc/interrupts
2164 This is used to record the number of interrupts per CPU per IO
2165 device. Since Linux 2.6.24, for the i386 and x86-64 architec‐
2166 tures, at least, this also includes interrupts internal to the
2167 system (that is, not associated with a device as such), such as
2168 NMI (nonmaskable interrupt), LOC (local timer interrupt), and
2169 for SMP systems, TLB (TLB flush interrupt), RES (rescheduling
2170 interrupt), CAL (remote function call interrupt), and possibly
2171 others. Very easy to read formatting, done in ASCII.
2172 /proc/iomem
2174 I/O memory map in Linux 2.4.
2175 /proc/ioports
2177 This is a list of currently registered Input-Output port regions
2178 that are in use.
2179 /proc/kallsyms (since Linux 2.5.71)
2181 This holds the kernel exported symbol definitions used by the
2182 modules(X) tools to dynamically link and bind loadable modules.
2183 In Linux 2.5.47 and earlier, a similar file with slightly dif‐
2184 ferent syntax was named ksyms.
2185 /proc/kcore
2187 This file represents the physical memory of the system and is
2188 stored in the ELF core file format. With this pseudo-file, and
2189 an unstripped kernel (/usr/src/linux/vmlinux) binary, GDB can be
2190 used to examine the current state of any kernel data structures.
2191 The total length of the file is the size of physical memory
2193 (RAM) plus 4 KiB.
2194 /proc/keys (since Linux 2.6.10)
2196 See keyrings(7).
2197 /proc/key-users (since Linux 2.6.10)
2199 See keyrings(7).
2200 /proc/kmsg
2202 This file can be used instead of the syslog(2) system call to
2203 read kernel messages. A process must have superuser privileges
2204 to read this file, and only one process should read this file.
2205 This file should not be read if a syslog process is running
2206 which uses the syslog(2) system call facility to log kernel mes‐
2207 sages.
2208 Information in this file is retrieved with the dmesg(1) program.
2210 /proc/kpagecgroup (since Linux 4.3)
2212 This file contains a 64-bit inode number of the memory cgroup
2213 each page is charged to, indexed by page frame number (see the
2214 discussion of /proc/pid/pagemap).
2215 The /proc/kpagecgroup file is present only if the CONFIG_MEMCG
2217 kernel configuration option is enabled.
2218 /proc/kpagecount (since Linux 2.6.25)
2220 This file contains a 64-bit count of the number of times each
2221 physical page frame is mapped, indexed by page frame number (see
2222 the discussion of /proc/pid/pagemap).
2223 The /proc/kpagecount file is present only if the CON‐
2225 FIG_PROC_PAGE_MONITOR kernel configuration option is enabled.
2226 /proc/kpageflags (since Linux 2.6.25)
2228 This file contains 64-bit masks corresponding to each physical
2229 page frame; it is indexed by page frame number (see the discus‐
2230 sion of /proc/pid/pagemap). The bits are as follows:
2231 0 - KPF_LOCKED
2233 1 - KPF_ERROR
2234 2 - KPF_REFERENCED
2235 3 - KPF_UPTODATE
2236 4 - KPF_DIRTY
2237 5 - KPF_LRU
2238 6 - KPF_ACTIVE
2239 7 - KPF_SLAB
2240 8 - KPF_WRITEBACK
2241 9 - KPF_RECLAIM
2242 10 - KPF_BUDDY
2243 11 - KPF_MMAP (since Linux 2.6.31)
2244 12 - KPF_ANON (since Linux 2.6.31)
2245 13 - KPF_SWAPCACHE (since Linux 2.6.31)
2246 14 - KPF_SWAPBACKED (since Linux 2.6.31)
2247 15 - KPF_COMPOUND_HEAD (since Linux 2.6.31)
2248 16 - KPF_COMPOUND_TAIL (since Linux 2.6.31)
2249 17 - KPF_HUGE (since Linux 2.6.31)
2250 18 - KPF_UNEVICTABLE (since Linux 2.6.31)
2251 19 - KPF_HWPOISON (since Linux 2.6.31)
2252 20 - KPF_NOPAGE (since Linux 2.6.31)
2253 21 - KPF_KSM (since Linux 2.6.32)
2254 22 - KPF_THP (since Linux 3.4)
2255 23 - KPF_BALLOON (since Linux 3.18)
2256 24 - KPF_ZERO_PAGE (since Linux 4.0)
2257 25 - KPF_IDLE (since Linux 4.3)
2258 26 - KPF_PGTABLE (since Linux 4.18)
2259 For further details on the meanings of these bits, see the ker‐
2261 nel source file Documentation/admin-guide/mm/pagemap.rst. Be‐
2262 fore Linux 2.6.29, KPF_WRITEBACK, KPF_RECLAIM, KPF_BUDDY, and
2263 KPF_LOCKED did not report correctly.
2264 The /proc/kpageflags file is present only if the CON‐
2266 FIG_PROC_PAGE_MONITOR kernel configuration option is enabled.
2267 /proc/ksyms (Linux 1.1.23–2.5.47)
2269 See /proc/kallsyms.
2270 /proc/loadavg
2272 The first three fields in this file are load average figures
2273 giving the number of jobs in the run queue (state R) or waiting
2274 for disk I/O (state D) averaged over 1, 5, and 15 minutes. They
2275 are the same as the load average numbers given by uptime(1) and
2276 other programs. The fourth field consists of two numbers sepa‐
2277 rated by a slash (/). The first of these is the number of cur‐
2278 rently runnable kernel scheduling entities (processes, threads).
2279 The value after the slash is the number of kernel scheduling en‐
2280 tities that currently exist on the system. The fifth field is
2281 the PID of the process that was most recently created on the
2282 system.
2283 /proc/locks
2285 This file shows current file locks (flock(2) and fcntl(2)) and
2286 leases (fcntl(2)).
2287 An example of the content shown in this file is the following:
2289 1: POSIX ADVISORY READ 5433 08:01:7864448 128 128
2291 2: FLOCK ADVISORY WRITE 2001 08:01:7864554 0 EOF
2292 3: FLOCK ADVISORY WRITE 1568 00:2f:32388 0 EOF
2293 4: POSIX ADVISORY WRITE 699 00:16:28457 0 EOF
2294 5: POSIX ADVISORY WRITE 764 00:16:21448 0 0
2295 6: POSIX ADVISORY READ 3548 08:01:7867240 1 1
2296 7: POSIX ADVISORY READ 3548 08:01:7865567 1826 2335
2297 8: OFDLCK ADVISORY WRITE -1 08:01:8713209 128 191
2298 The fields shown in each line are as follows:
2300 [1] The ordinal position of the lock in the list.
2302 [2] The lock type. Values that may appear here include:
2304 FLOCK This is a BSD file lock created using flock(2).
2306 OFDLCK This is an open file description (OFD) lock created
2308 using fcntl(2).
2309 POSIX This is a POSIX byte-range lock created using fc‐
2311 ntl(2).
2312 [3] Among the strings that can appear here are the following:
2314 ADVISORY
2316 This is an advisory lock.
2317 MANDATORY
2319 This is a mandatory lock.
2320 [4] The type of lock. Values that can appear here are:
2322 READ This is a POSIX or OFD read lock, or a BSD shared
2324 lock.
2325 WRITE This is a POSIX or OFD write lock, or a BSD exclu‐
2327 sive lock.
2328 [5] The PID of the process that owns the lock.
2330 Because OFD locks are not owned by a single process (since
2332 multiple processes may have file descriptors that refer to
2333 the same open file description), the value -1 is displayed
2334 in this field for OFD locks. (Before Linux 4.14, a bug
2335 meant that the PID of the process that initially acquired
2336 the lock was displayed instead of the value -1.)
2337 [6] Three colon-separated subfields that identify the major and
2339 minor device ID of the device containing the filesystem
2340 where the locked file resides, followed by the inode number
2341 of the locked file.
2342 [7] The byte offset of the first byte of the lock. For BSD
2344 locks, this value is always 0.
2345 [8] The byte offset of the last byte of the lock. EOF in this
2347 field means that the lock extends to the end of the file.
2348 For BSD locks, the value shown is always EOF.
2349 Since Linux 4.9, the list of locks shown in /proc/locks is fil‐
2351 tered to show just the locks for the processes in the PID name‐
2352 space (see pid_namespaces(7)) for which the /proc filesystem was
2353 mounted. (In the initial PID namespace, there is no filtering
2354 of the records shown in this file.)
2355 The lslocks(8) command provides a bit more information about
2357 each lock.
2358 /proc/malloc (only up to and including Linux 2.2)
2360 This file is present only if CONFIG_DEBUG_MALLOC was defined
2361 during compilation.
2362 /proc/meminfo
2364 This file reports statistics about memory usage on the system.
2365 It is used by free(1) to report the amount of free and used mem‐
2366 ory (both physical and swap) on the system as well as the shared
2367 memory and buffers used by the kernel. Each line of the file
2368 consists of a parameter name, followed by a colon, the value of
2369 the parameter, and an option unit of measurement (e.g., "kB").
2370 The list below describes the parameter names and the format
2371 specifier required to read the field value. Except as noted be‐
2372 low, all of the fields have been present since at least Linux
2373 2.6.0. Some fields are displayed only if the kernel was config‐
2374 ured with various options; those dependencies are noted in the
2375 list.
2376 MemTotal %lu
2378 Total usable RAM (i.e., physical RAM minus a few reserved
2379 bits and the kernel binary code).
2380 MemFree %lu
2382 The sum of LowFree+HighFree.
2383 MemAvailable %lu (since Linux 3.14)
2385 An estimate of how much memory is available for starting
2386 new applications, without swapping.
2387 Buffers %lu
2389 Relatively temporary storage for raw disk blocks that
2390 shouldn't get tremendously large (20 MB or so).
2391 Cached %lu
2393 In-memory cache for files read from the disk (the page
2394 cache). Doesn't include SwapCached.
2395 SwapCached %lu
2397 Memory that once was swapped out, is swapped back in but
2398 still also is in the swap file. (If memory pressure is
2399 high, these pages don't need to be swapped out again be‐
2400 cause they are already in the swap file. This saves
2401 I/O.)
2402 Active %lu
2404 Memory that has been used more recently and usually not
2405 reclaimed unless absolutely necessary.
2406 Inactive %lu
2408 Memory which has been less recently used. It is more el‐
2409 igible to be reclaimed for other purposes.
2410 Active(anon) %lu (since Linux 2.6.28)
2412 [To be documented.]
2413 Inactive(anon) %lu (since Linux 2.6.28)
2415 [To be documented.]
2416 Active(file) %lu (since Linux 2.6.28)
2418 [To be documented.]
2419 Inactive(file) %lu (since Linux 2.6.28)
2421 [To be documented.]
2422 Unevictable %lu (since Linux 2.6.28)
2424 (From Linux 2.6.28 to Linux 2.6.30, CONFIG_UN‐
2425 EVICTABLE_LRU was required.) [To be documented.]
2426 Mlocked %lu (since Linux 2.6.28)
2428 (From Linux 2.6.28 to Linux 2.6.30, CONFIG_UN‐
2429 EVICTABLE_LRU was required.) [To be documented.]
2430 HighTotal %lu
2432 (Starting with Linux 2.6.19, CONFIG_HIGHMEM is required.)
2433 Total amount of highmem. Highmem is all memory above
2434 ~860 MB of physical memory. Highmem areas are for use by
2435 user-space programs, or for the page cache. The kernel
2436 must use tricks to access this memory, making it slower
2437 to access than lowmem.
2438 HighFree %lu
2440 (Starting with Linux 2.6.19, CONFIG_HIGHMEM is required.)
2441 Amount of free highmem.
2442 LowTotal %lu
2444 (Starting with Linux 2.6.19, CONFIG_HIGHMEM is required.)
2445 Total amount of lowmem. Lowmem is memory which can be
2446 used for everything that highmem can be used for, but it
2447 is also available for the kernel's use for its own data
2448 structures. Among many other things, it is where every‐
2449 thing from Slab is allocated. Bad things happen when
2450 you're out of lowmem.
2451 LowFree %lu
2453 (Starting with Linux 2.6.19, CONFIG_HIGHMEM is required.)
2454 Amount of free lowmem.
2455 MmapCopy %lu (since Linux 2.6.29)
2457 (CONFIG_MMU is required.) [To be documented.]
2458 SwapTotal %lu
2460 Total amount of swap space available.
2461 SwapFree %lu
2463 Amount of swap space that is currently unused.
2464 Dirty %lu
2466 Memory which is waiting to get written back to the disk.
2467 Writeback %lu
2469 Memory which is actively being written back to the disk.
2470 AnonPages %lu (since Linux 2.6.18)
2472 Non-file backed pages mapped into user-space page tables.
2473 Mapped %lu
2475 Files which have been mapped into memory (with mmap(2)),
2476 such as libraries.
2477 Shmem %lu (since Linux 2.6.32)
2479 Amount of memory consumed in tmpfs(5) filesystems.
2480 KReclaimable %lu (since Linux 4.20)
2482 Kernel allocations that the kernel will attempt to re‐
2483 claim under memory pressure. Includes SReclaimable (be‐
2484 low), and other direct allocations with a shrinker.
2485 Slab %lu
2487 In-kernel data structures cache. (See slabinfo(5).)
2488 SReclaimable %lu (since Linux 2.6.19)
2490 Part of Slab, that might be reclaimed, such as caches.
2491 SUnreclaim %lu (since Linux 2.6.19)
2493 Part of Slab, that cannot be reclaimed on memory pres‐
2494 sure.
2495 KernelStack %lu (since Linux 2.6.32)
2497 Amount of memory allocated to kernel stacks.
2498 PageTables %lu (since Linux 2.6.18)
2500 Amount of memory dedicated to the lowest level of page
2501 tables.
2502 Quicklists %lu (since Linux 2.6.27)
2504 (CONFIG_QUICKLIST is required.) [To be documented.]
2505 NFS_Unstable %lu (since Linux 2.6.18)
2507 NFS pages sent to the server, but not yet committed to
2508 stable storage.
2509 Bounce %lu (since Linux 2.6.18)
2511 Memory used for block device "bounce buffers".
2512 WritebackTmp %lu (since Linux 2.6.26)
2514 Memory used by FUSE for temporary writeback buffers.
2515 CommitLimit %lu (since Linux 2.6.10)
2517 This is the total amount of memory currently available to
2518 be allocated on the system, expressed in kilobytes. This
2519 limit is adhered to only if strict overcommit accounting
2520 is enabled (mode 2 in /proc/sys/vm/overcommit_memory).
2521 The limit is calculated according to the formula de‐
2522 scribed under /proc/sys/vm/overcommit_memory. For fur‐
2523 ther details, see the kernel source file Documenta‐
2524 tion/vm/overcommit-accounting.rst.
2525 Committed_AS %lu
2527 The amount of memory presently allocated on the system.
2528 The committed memory is a sum of all of the memory which
2529 has been allocated by processes, even if it has not been
2530 "used" by them as of yet. A process which allocates 1 GB
2531 of memory (using malloc(3) or similar), but touches only
2532 300 MB of that memory will show up as using only 300 MB
2533 of memory even if it has the address space allocated for
2534 the entire 1 GB.
2535 This 1 GB is memory which has been "committed" to by the
2537 VM and can be used at any time by the allocating applica‐
2538 tion. With strict overcommit enabled on the system (mode
2539 2 in /proc/sys/vm/overcommit_memory), allocations which
2540 would exceed the CommitLimit will not be permitted. This
2541 is useful if one needs to guarantee that processes will
2542 not fail due to lack of memory once that memory has been
2543 successfully allocated.
2544 VmallocTotal %lu
2546 Total size of vmalloc memory area.
2547 VmallocUsed %lu
2549 Amount of vmalloc area which is used. Since Linux 4.4,
2550 this field is no longer calculated, and is hard coded as
2551 0. See /proc/vmallocinfo.
2552 VmallocChunk %lu
2554 Largest contiguous block of vmalloc area which is free.
2555 Since Linux 4.4, this field is no longer calculated and
2556 is hard coded as 0. See /proc/vmallocinfo.
2557 HardwareCorrupted %lu (since Linux 2.6.32)
2559 (CONFIG_MEMORY_FAILURE is required.) [To be documented.]
2560 LazyFree %lu (since Linux 4.12)
2562 Shows the amount of memory marked by madvise(2)
2563 MADV_FREE.
2564 AnonHugePages %lu (since Linux 2.6.38)
2566 (CONFIG_TRANSPARENT_HUGEPAGE is required.) Non-file
2567 backed huge pages mapped into user-space page tables.
2568 ShmemHugePages %lu (since Linux 4.8)
2570 (CONFIG_TRANSPARENT_HUGEPAGE is required.) Memory used
2571 by shared memory (shmem) and tmpfs(5) allocated with huge
2572 pages.
2573 ShmemPmdMapped %lu (since Linux 4.8)
2575 (CONFIG_TRANSPARENT_HUGEPAGE is required.) Shared memory
2576 mapped into user space with huge pages.
2577 CmaTotal %lu (since Linux 3.1)
2579 Total CMA (Contiguous Memory Allocator) pages. (CON‐
2580 FIG_CMA is required.)
2581 CmaFree %lu (since Linux 3.1)
2583 Free CMA (Contiguous Memory Allocator) pages. (CON‐
2584 FIG_CMA is required.)
2585 HugePages_Total %lu
2587 (CONFIG_HUGETLB_PAGE is required.) The size of the pool
2588 of huge pages.
2589 HugePages_Free %lu
2591 (CONFIG_HUGETLB_PAGE is required.) The number of huge
2592 pages in the pool that are not yet allocated.
2593 HugePages_Rsvd %lu (since Linux 2.6.17)
2595 (CONFIG_HUGETLB_PAGE is required.) This is the number of
2596 huge pages for which a commitment to allocate from the
2597 pool has been made, but no allocation has yet been made.
2598 These reserved huge pages guarantee that an application
2599 will be able to allocate a huge page from the pool of
2600 huge pages at fault time.
2601 HugePages_Surp %lu (since Linux 2.6.24)
2603 (CONFIG_HUGETLB_PAGE is required.) This is the number of
2604 huge pages in the pool above the value in
2605 /proc/sys/vm/nr_hugepages. The maximum number of surplus
2606 huge pages is controlled by /proc/sys/vm/nr_overcom‐
2607 mit_hugepages.
2608 Hugepagesize %lu
2610 (CONFIG_HUGETLB_PAGE is required.) The size of huge
2611 pages.
2612 DirectMap4k %lu (since Linux 2.6.27)
2614 Number of bytes of RAM linearly mapped by kernel in 4 kB
2615 pages. (x86.)
2616 DirectMap4M %lu (since Linux 2.6.27)
2618 Number of bytes of RAM linearly mapped by kernel in 4 MB
2619 pages. (x86 with CONFIG_X86_64 or CONFIG_X86_PAE en‐
2620 abled.)
2621 DirectMap2M %lu (since Linux 2.6.27)
2623 Number of bytes of RAM linearly mapped by kernel in 2 MB
2624 pages. (x86 with neither CONFIG_X86_64 nor CON‐
2625 FIG_X86_PAE enabled.)
2626 DirectMap1G %lu (since Linux 2.6.27)
2628 (x86 with CONFIG_X86_64 and CONFIG_X86_DIRECT_GBPAGES en‐
2629 abled.)
2630 /proc/modules
2632 A text list of the modules that have been loaded by the system.
2633 See also lsmod(8).
2634 /proc/mounts
2636 Before Linux 2.4.19, this file was a list of all the filesystems
2637 currently mounted on the system. With the introduction of per-
2638 process mount namespaces in Linux 2.4.19 (see mount_name‐
2639 spaces(7)), this file became a link to /proc/self/mounts, which
2640 lists the mounts of the process's own mount namespace. The for‐
2641 mat of this file is documented in fstab(5).
2642 /proc/mtrr
2644 Memory Type Range Registers. See the Linux kernel source file
2645 Documentation/x86/mtrr.rst (or Documentation/x86/mtrr.txt before
2646 Linux 5.2, or Documentation/mtrr.txt before Linux 2.6.28) for
2647 details.
2648 /proc/net
2650 This directory contains various files and subdirectories con‐
2651 taining information about the networking layer. The files con‐
2652 tain ASCII structures and are, therefore, readable with cat(1).
2653 However, the standard netstat(8) suite provides much cleaner ac‐
2654 cess to these files.
2655 With the advent of network namespaces, various information re‐
2657 lating to the network stack is virtualized (see network_name‐
2658 spaces(7)). Thus, since Linux 2.6.25, /proc/net is a symbolic
2659 link to the directory /proc/self/net, which contains the same
2660 files and directories as listed below. However, these files and
2661 directories now expose information for the network namespace of
2662 which the process is a member.
2663 /proc/net/arp
2665 This holds an ASCII readable dump of the kernel ARP table used
2666 for address resolutions. It will show both dynamically learned
2667 and preprogrammed ARP entries. The format is:
2668 IP address HW type Flags HW address Mask Device
2670 192.168.0.50 0x1 0x2 00:50:BF:25:68:F3 * eth0
2671 192.168.0.250 0x1 0xc 00:00:00:00:00:00 * eth0
2672 Here "IP address" is the IPv4 address of the machine and the "HW
2674 type" is the hardware type of the address from RFC 826. The
2675 flags are the internal flags of the ARP structure (as defined in
2676 /usr/include/linux/if_arp.h) and the "HW address" is the data
2677 link layer mapping for that IP address if it is known.
2678 /proc/net/dev
2680 The dev pseudo-file contains network device status information.
2681 This gives the number of received and sent packets, the number
2682 of errors and collisions and other basic statistics. These are
2683 used by the ifconfig(8) program to report device status. The
2684 format is:
2685 Inter-| Receive | Transmit
2687 face |bytes packets errs drop fifo frame compressed multicast|bytes packets errs drop fifo colls carrier compressed
2688 lo: 2776770 11307 0 0 0 0 0 0 2776770 11307 0 0 0 0 0 0
2689 eth0: 1215645 2751 0 0 0 0 0 0 1782404 4324 0 0 0 427 0 0
2690 ppp0: 1622270 5552 1 0 0 0 0 0 354130 5669 0 0 0 0 0 0
2691 tap0: 7714 81 0 0 0 0 0 0 7714 81 0 0 0 0 0 0
2692 /proc/net/dev_mcast
2694 Defined in /usr/src/linux/net/core/dev_mcast.c:
2695 indx interface_name dmi_u dmi_g dmi_address
2697 2 eth0 1 0 01005e000001
2698 3 eth1 1 0 01005e000001
2699 4 eth2 1 0 01005e000001
2700 /proc/net/igmp
2702 Internet Group Management Protocol. Defined in
2703 /usr/src/linux/net/core/igmp.c.
2704 /proc/net/rarp
2706 This file uses the same format as the arp file and contains the
2707 current reverse mapping database used to provide rarp(8) reverse
2708 address lookup services. If RARP is not configured into the
2709 kernel, this file will not be present.
2710 /proc/net/raw
2712 Holds a dump of the RAW socket table. Much of the information
2713 is not of use apart from debugging. The "sl" value is the ker‐
2714 nel hash slot for the socket, the "local_address" is the local
2715 address and protocol number pair. "St" is the internal status
2716 of the socket. The "tx_queue" and "rx_queue" are the outgoing
2717 and incoming data queue in terms of kernel memory usage. The
2718 "tr", "tm->when", and "rexmits" fields are not used by RAW. The
2719 "uid" field holds the effective UID of the creator of the
2720 socket.
2721 /proc/net/snmp
2723 This file holds the ASCII data needed for the IP, ICMP, TCP, and
2724 UDP management information bases for an SNMP agent.
2725 /proc/net/tcp
2727 Holds a dump of the TCP socket table. Much of the information
2728 is not of use apart from debugging. The "sl" value is the ker‐
2729 nel hash slot for the socket, the "local_address" is the local
2730 address and port number pair. The "rem_address" is the remote
2731 address and port number pair (if connected). "St" is the inter‐
2732 nal status of the socket. The "tx_queue" and "rx_queue" are the
2733 outgoing and incoming data queue in terms of kernel memory us‐
2734 age. The "tr", "tm->when", and "rexmits" fields hold internal
2735 information of the kernel socket state and are useful only for
2736 debugging. The "uid" field holds the effective UID of the cre‐
2737 ator of the socket.
2738 /proc/net/udp
2740 Holds a dump of the UDP socket table. Much of the information
2741 is not of use apart from debugging. The "sl" value is the ker‐
2742 nel hash slot for the socket, the "local_address" is the local
2743 address and port number pair. The "rem_address" is the remote
2744 address and port number pair (if connected). "St" is the inter‐
2745 nal status of the socket. The "tx_queue" and "rx_queue" are the
2746 outgoing and incoming data queue in terms of kernel memory us‐
2747 age. The "tr", "tm->when", and "rexmits" fields are not used by
2748 UDP. The "uid" field holds the effective UID of the creator of
2749 the socket. The format is:
2750 sl local_address rem_address st tx_queue rx_queue tr rexmits tm->when uid
2752 1: 01642C89:0201 0C642C89:03FF 01 00000000:00000001 01:000071BA 00000000 0
2753 1: 00000000:0801 00000000:0000 0A 00000000:00000000 00:00000000 6F000100 0
2754 1: 00000000:0201 00000000:0000 0A 00000000:00000000 00:00000000 00000000 0
2755 /proc/net/unix
2757 Lists the UNIX domain sockets present within the system and
2758 their status. The format is:
2759 Num RefCount Protocol Flags Type St Inode Path
2761 0: 00000002 00000000 00000000 0001 03 42
2762 1: 00000001 00000000 00010000 0001 01 1948 /dev/printer
2763 The fields are as follows:
2765 Num: the kernel table slot number.
2767 RefCount: the number of users of the socket.
2769 Protocol: currently always 0.
2771 Flags: the internal kernel flags holding the status of the
2773 socket.
2774 Type: the socket type. For SOCK_STREAM sockets, this is
2776 0001; for SOCK_DGRAM sockets, it is 0002; and for
2777 SOCK_SEQPACKET sockets, it is 0005.
2778 St: the internal state of the socket.
2780 Inode: the inode number of the socket.
2782 Path: the bound pathname (if any) of the socket. Sockets in
2784 the abstract namespace are included in the list, and
2785 are shown with a Path that commences with the charac‐
2786 ter '@'.
2787 /proc/net/netfilter/nfnetlink_queue
2789 This file contains information about netfilter user-space queue‐
2790 ing, if used. Each line represents a queue. Queues that have
2791 not been subscribed to by user space are not shown.
2792 1 4207 0 2 65535 0 0 0 1
2794 (1) (2) (3)(4) (5) (6) (7) (8)
2795 The fields in each line are:
2797 (1) The ID of the queue. This matches what is specified in the
2799 --queue-num or --queue-balance options to the iptables(8)
2800 NFQUEUE target. See iptables-extensions(8) for more infor‐
2801 mation.
2802 (2) The netlink port ID subscribed to the queue.
2804 (3) The number of packets currently queued and waiting to be
2806 processed by the application.
2807 (4) The copy mode of the queue. It is either 1 (metadata only)
2809 or 2 (also copy payload data to user space).
2810 (5) Copy range; that is, how many bytes of packet payload
2812 should be copied to user space at most.
2813 (6) queue dropped. Number of packets that had to be dropped by
2815 the kernel because too many packets are already waiting for
2816 user space to send back the mandatory accept/drop verdicts.
2817 (7) queue user dropped. Number of packets that were dropped
2819 within the netlink subsystem. Such drops usually happen
2820 when the corresponding socket buffer is full; that is, user
2821 space is not able to read messages fast enough.
2822 (8) sequence number. Every queued packet is associated with a
2824 (32-bit) monotonically increasing sequence number. This
2825 shows the ID of the most recent packet queued.
2826 The last number exists only for compatibility reasons and is al‐
2828 ways 1.
2829 /proc/partitions
2831 Contains the major and minor numbers of each partition as well
2832 as the number of 1024-byte blocks and the partition name.
2833 /proc/pci
2835 This is a listing of all PCI devices found during kernel ini‐
2836 tialization and their configuration.
2837 This file has been deprecated in favor of a new /proc interface
2839 for PCI (/proc/bus/pci). It became optional in Linux 2.2
2840 (available with CONFIG_PCI_OLD_PROC set at kernel compilation).
2841 It became once more nonoptionally enabled in Linux 2.4. Next,
2842 it was deprecated in Linux 2.6 (still available with CON‐
2843 FIG_PCI_LEGACY_PROC set), and finally removed altogether since
2844 Linux 2.6.17.
2845 /proc/profile (since Linux 2.4)
2847 This file is present only if the kernel was booted with the pro‐
2848 file=1 command-line option. It exposes kernel profiling infor‐
2849 mation in a binary format for use by readprofile(1). Writing
2850 (e.g., an empty string) to this file resets the profiling coun‐
2851 ters; on some architectures, writing a binary integer "profiling
2852 multiplier" of size sizeof(int) sets the profiling interrupt
2853 frequency.
2854 /proc/scsi
2856 A directory with the scsi mid-level pseudo-file and various SCSI
2857 low-level driver directories, which contain a file for each SCSI
2858 host in this system, all of which give the status of some part
2859 of the SCSI IO subsystem. These files contain ASCII structures
2860 and are, therefore, readable with cat(1).
2861 You can also write to some of the files to reconfigure the sub‐
2863 system or switch certain features on or off.
2864 /proc/scsi/scsi
2866 This is a listing of all SCSI devices known to the kernel. The
2867 listing is similar to the one seen during bootup. scsi cur‐
2868 rently supports only the add-single-device command which allows
2869 root to add a hotplugged device to the list of known devices.
2870 The command
2872 echo 'scsi add-single-device 1 0 5 0' > /proc/scsi/scsi
2874 will cause host scsi1 to scan on SCSI channel 0 for a device on
2876 ID 5 LUN 0. If there is already a device known on this address
2877 or the address is invalid, an error will be returned.
2878 /proc/scsi/drivername
2880 drivername can currently be NCR53c7xx, aha152x, aha1542,
2881 aha1740, aic7xxx, buslogic, eata_dma, eata_pio, fdomain, in2000,
2882 pas16, qlogic, scsi_debug, seagate, t128, u15-24f, ultrastore,
2883 or wd7000. These directories show up for all drivers that reg‐
2884 istered at least one SCSI HBA. Every directory contains one
2885 file per registered host. Every host-file is named after the
2886 number the host was assigned during initialization.
2887 Reading these files will usually show driver and host configura‐
2889 tion, statistics, and so on.
2890 Writing to these files allows different things on different
2892 hosts. For example, with the latency and nolatency commands,
2893 root can switch on and off command latency measurement code in
2894 the eata_dma driver. With the lockup and unlock commands, root
2895 can control bus lockups simulated by the scsi_debug driver.
2896 /proc/self
2898 This directory refers to the process accessing the /proc
2899 filesystem, and is identical to the /proc directory named by the
2900 process ID of the same process.
2901 /proc/slabinfo
2903 Information about kernel caches. See slabinfo(5) for details.
2904 /proc/stat
2906 kernel/system statistics. Varies with architecture. Common en‐
2907 tries include:
2908 cpu 10132153 290696 3084719 46828483 16683 0 25195 0 175628 0
2910 cpu0 1393280 32966 572056 13343292 6130 0 17875 0 23933 0
2911 The amount of time, measured in units of USER_HZ
2912 (1/100ths of a second on most architectures, use
2913 sysconf(_SC_CLK_TCK) to obtain the right value), that the
2914 system ("cpu" line) or the specific CPU ("cpuN" line)
2915 spent in various states:
2916 user (1) Time spent in user mode.
2918 nice (2) Time spent in user mode with low priority
2920 (nice).
2921 system (3) Time spent in system mode.
2923 idle (4) Time spent in the idle task. This value
2925 should be USER_HZ times the second entry in the
2926 /proc/uptime pseudo-file.
2927 iowait (since Linux 2.5.41)
2929 (5) Time waiting for I/O to complete. This value
2930 is not reliable, for the following reasons:
2931 • The CPU will not wait for I/O to complete;
2933 iowait is the time that a task is waiting for
2934 I/O to complete. When a CPU goes into idle
2935 state for outstanding task I/O, another task
2936 will be scheduled on this CPU.
2937 • On a multi-core CPU, the task waiting for I/O
2939 to complete is not running on any CPU, so the
2940 iowait of each CPU is difficult to calculate.
2941 • The value in this field may decrease in certain
2943 conditions.
2944 irq (since Linux 2.6.0)
2946 (6) Time servicing interrupts.
2947 softirq (since Linux 2.6.0)
2949 (7) Time servicing softirqs.
2950 steal (since Linux 2.6.11)
2952 (8) Stolen time, which is the time spent in other
2953 operating systems when running in a virtualized
2954 environment
2955 guest (since Linux 2.6.24)
2957 (9) Time spent running a virtual CPU for guest op‐
2958 erating systems under the control of the Linux
2959 kernel.
2960 guest_nice (since Linux 2.6.33)
2962 (10) Time spent running a niced guest (virtual CPU
2963 for guest operating systems under the control of
2964 the Linux kernel).
2965 page 5741 1808
2967 The number of pages the system paged in and the number
2968 that were paged out (from disk).
2969 swap 1 0
2971 The number of swap pages that have been brought in and
2972 out.
2973 intr 1462898
2975 This line shows counts of interrupts serviced since boot
2976 time, for each of the possible system interrupts. The
2977 first column is the total of all interrupts serviced in‐
2978 cluding unnumbered architecture specific interrupts; each
2979 subsequent column is the total for that particular num‐
2980 bered interrupt. Unnumbered interrupts are not shown,
2981 only summed into the total.
2982 disk_io: (2,0):(31,30,5764,1,2) (3,0):...
2984 (major,disk_idx):(noinfo, read_io_ops, blks_read,
2985 write_io_ops, blks_written)
2986 (Linux 2.4 only)
2987 ctxt 115315
2989 The number of context switches that the system underwent.
2990 btime 769041601
2992 boot time, in seconds since the Epoch, 1970-01-01
2993 00:00:00 +0000 (UTC).
2994 processes 86031
2996 Number of forks since boot.
2997 procs_running 6
2999 Number of processes in runnable state. (Linux 2.5.45 on‐
3000 ward.)
3001 procs_blocked 2
3003 Number of processes blocked waiting for I/O to complete.
3004 (Linux 2.5.45 onward.)
3005 softirq 229245889 94 60001584 13619 5175704 2471304 28 51212741
3007 59130143 0 51240672
3008 This line shows the number of softirq for all CPUs. The
3009 first column is the total of all softirqs and each subse‐
3010 quent column is the total for particular softirq. (Linux
3011 2.6.31 onward.)
3012 /proc/swaps
3014 Swap areas in use. See also swapon(8).
3015 /proc/sys
3017 This directory (present since Linux 1.3.57) contains a number of
3018 files and subdirectories corresponding to kernel variables.
3019 These variables can be read and in some cases modified using the
3020 /proc filesystem, and the (deprecated) sysctl(2) system call.
3021 String values may be terminated by either '\0' or '\n'.
3023 Integer and long values may be written either in decimal or in
3025 hexadecimal notation (e.g., 0x3FFF). When writing multiple in‐
3026 teger or long values, these may be separated by any of the fol‐
3027 lowing whitespace characters: ' ', '\t', or '\n'. Using other
3028 separators leads to the error EINVAL.
3029 /proc/sys/abi (since Linux 2.4.10)
3031 This directory may contain files with application binary infor‐
3032 mation. See the Linux kernel source file Documenta‐
3033 tion/sysctl/abi.rst (or Documentation/sysctl/abi.txt before
3034 Linux 5.3) for more information.
3035 /proc/sys/debug
3037 This directory may be empty.
3038 /proc/sys/dev
3040 This directory contains device-specific information (e.g.,
3041 dev/cdrom/info). On some systems, it may be empty.
3042 /proc/sys/fs
3044 This directory contains the files and subdirectories for kernel
3045 variables related to filesystems.
3046 /proc/sys/fs/aio-max-nr and /proc/sys/fs/aio-nr (since Linux 2.6.4)
3048 aio-nr is the running total of the number of events specified by
3049 io_setup(2) calls for all currently active AIO contexts. If
3050 aio-nr reaches aio-max-nr, then io_setup(2) will fail with the
3051 error EAGAIN. Raising aio-max-nr does not result in the preal‐
3052 location or resizing of any kernel data structures.
3053 /proc/sys/fs/binfmt_misc
3055 Documentation for files in this directory can be found in the
3056 Linux kernel source in the file Documentation/ad‐
3057 min-guide/binfmt-misc.rst (or in Documentation/binfmt_misc.txt
3058 on older kernels).
3059 /proc/sys/fs/dentry-state (since Linux 2.2)
3061 This file contains information about the status of the directory
3062 cache (dcache). The file contains six numbers, nr_dentry,
3063 nr_unused, age_limit (age in seconds), want_pages (pages re‐
3064 quested by system) and two dummy values.
3065 • nr_dentry is the number of allocated dentries (dcache en‐
3067 tries). This field is unused in Linux 2.2.
3068 • nr_unused is the number of unused dentries.
3070 • age_limit is the age in seconds after which dcache entries
3072 can be reclaimed when memory is short.
3073 • want_pages is nonzero when the kernel has called
3075 shrink_dcache_pages() and the dcache isn't pruned yet.
3076 /proc/sys/fs/dir-notify-enable
3078 This file can be used to disable or enable the dnotify interface
3079 described in fcntl(2) on a system-wide basis. A value of 0 in
3080 this file disables the interface, and a value of 1 enables it.
3081 /proc/sys/fs/dquot-max
3083 This file shows the maximum number of cached disk quota entries.
3084 On some (2.4) systems, it is not present. If the number of free
3085 cached disk quota entries is very low and you have some awesome
3086 number of simultaneous system users, you might want to raise the
3087 limit.
3088 /proc/sys/fs/dquot-nr
3090 This file shows the number of allocated disk quota entries and
3091 the number of free disk quota entries.
3092 /proc/sys/fs/epoll (since Linux 2.6.28)
3094 This directory contains the file max_user_watches, which can be
3095 used to limit the amount of kernel memory consumed by the epoll
3096 interface. For further details, see epoll(7).
3097 /proc/sys/fs/file-max
3099 This file defines a system-wide limit on the number of open
3100 files for all processes. System calls that fail when encounter‐
3101 ing this limit fail with the error ENFILE. (See also setr‐
3102 limit(2), which can be used by a process to set the per-process
3103 limit, RLIMIT_NOFILE, on the number of files it may open.) If
3104 you get lots of error messages in the kernel log about running
3105 out of file handles (open file descriptions) (look for "VFS:
3106 file-max limit <number> reached"), try increasing this value:
3107 echo 100000 > /proc/sys/fs/file-max
3109 Privileged processes (CAP_SYS_ADMIN) can override the file-max
3111 limit.
3112 /proc/sys/fs/file-nr
3114 This (read-only) file contains three numbers: the number of al‐
3115 located file handles (i.e., the number of open file descrip‐
3116 tions; see open(2)); the number of free file handles; and the
3117 maximum number of file handles (i.e., the same value as
3118 /proc/sys/fs/file-max). If the number of allocated file handles
3119 is close to the maximum, you should consider increasing the max‐
3120 imum. Before Linux 2.6, the kernel allocated file handles dy‐
3121 namically, but it didn't free them again. Instead the free file
3122 handles were kept in a list for reallocation; the "free file
3123 handles" value indicates the size of that list. A large number
3124 of free file handles indicates that there was a past peak in the
3125 usage of open file handles. Since Linux 2.6, the kernel does
3126 deallocate freed file handles, and the "free file handles" value
3127 is always zero.
3128 /proc/sys/fs/inode-max (only present until Linux 2.2)
3130 This file contains the maximum number of in-memory inodes. This
3131 value should be 3–4 times larger than the value in file-max,
3132 since stdin, stdout and network sockets also need an inode to
3133 handle them. When you regularly run out of inodes, you need to
3134 increase this value.
3135 Starting with Linux 2.4, there is no longer a static limit on
3137 the number of inodes, and this file is removed.
3138 /proc/sys/fs/inode-nr
3140 This file contains the first two values from inode-state.
3141 /proc/sys/fs/inode-state
3143 This file contains seven numbers: nr_inodes, nr_free_inodes,
3144 preshrink, and four dummy values (always zero).
3145 nr_inodes is the number of inodes the system has allocated.
3147 nr_free_inodes represents the number of free inodes.
3148 preshrink is nonzero when the nr_inodes > inode-max and the sys‐
3150 tem needs to prune the inode list instead of allocating more;
3151 since Linux 2.4, this field is a dummy value (always zero).
3152 /proc/sys/fs/inotify (since Linux 2.6.13)
3154 This directory contains files max_queued_events, max_user_in‐
3155 stances, and max_user_watches, that can be used to limit the
3156 amount of kernel memory consumed by the inotify interface. For
3157 further details, see inotify(7).
3158 /proc/sys/fs/lease-break-time
3160 This file specifies the grace period that the kernel grants to a
3161 process holding a file lease (fcntl(2)) after it has sent a sig‐
3162 nal to that process notifying it that another process is waiting
3163 to open the file. If the lease holder does not remove or down‐
3164 grade the lease within this grace period, the kernel forcibly
3165 breaks the lease.
3166 /proc/sys/fs/leases-enable
3168 This file can be used to enable or disable file leases (fc‐
3169 ntl(2)) on a system-wide basis. If this file contains the value
3170 0, leases are disabled. A nonzero value enables leases.
3171 /proc/sys/fs/mount-max (since Linux 4.9)
3173 The value in this file specifies the maximum number of mounts
3174 that may exist in a mount namespace. The default value in this
3175 file is 100,000.
3176 /proc/sys/fs/mqueue (since Linux 2.6.6)
3178 This directory contains files msg_max, msgsize_max, and
3179 queues_max, controlling the resources used by POSIX message
3180 queues. See mq_overview(7) for details.
3181 /proc/sys/fs/nr_open (since Linux 2.6.25)
3183 This file imposes a ceiling on the value to which the
3184 RLIMIT_NOFILE resource limit can be raised (see getrlimit(2)).
3185 This ceiling is enforced for both unprivileged and privileged
3186 process. The default value in this file is 1048576. (Before
3187 Linux 2.6.25, the ceiling for RLIMIT_NOFILE was hard-coded to
3188 the same value.)
3189 /proc/sys/fs/overflowgid and /proc/sys/fs/overflowuid
3191 These files allow you to change the value of the fixed UID and
3192 GID. The default is 65534. Some filesystems support only
3193 16-bit UIDs and GIDs, although in Linux UIDs and GIDs are 32
3194 bits. When one of these filesystems is mounted with writes en‐
3195 abled, any UID or GID that would exceed 65535 is translated to
3196 the overflow value before being written to disk.
3197 /proc/sys/fs/pipe-max-size (since Linux 2.6.35)
3199 See pipe(7).
3200 /proc/sys/fs/pipe-user-pages-hard (since Linux 4.5)
3202 See pipe(7).
3203 /proc/sys/fs/pipe-user-pages-soft (since Linux 4.5)
3205 See pipe(7).
3206 /proc/sys/fs/protected_fifos (since Linux 4.19)
3208 The value in this file is/can be set to one of the following:
3209 0 Writing to FIFOs is unrestricted.
3211 1 Don't allow O_CREAT open(2) on FIFOs that the caller doesn't
3213 own in world-writable sticky directories, unless the FIFO is
3214 owned by the owner of the directory.
3215 2 As for the value 1, but the restriction also applies to
3217 group-writable sticky directories.
3218 The intent of the above protections is to avoid unintentional
3220 writes to an attacker-controlled FIFO when a program expected to
3221 create a regular file.
3222 /proc/sys/fs/protected_hardlinks (since Linux 3.6)
3224 When the value in this file is 0, no restrictions are placed on
3225 the creation of hard links (i.e., this is the historical behav‐
3226 ior before Linux 3.6). When the value in this file is 1, a hard
3227 link can be created to a target file only if one of the follow‐
3228 ing conditions is true:
3229 • The calling process has the CAP_FOWNER capability in its user
3231 namespace and the file UID has a mapping in the namespace.
3232 • The filesystem UID of the process creating the link matches
3234 the owner (UID) of the target file (as described in creden‐
3235 tials(7), a process's filesystem UID is normally the same as
3236 its effective UID).
3237 • All of the following conditions are true:
3239 • the target is a regular file;
3241 • the target file does not have its set-user-ID mode bit
3243 enabled;
3244 • the target file does not have both its set-group-ID and
3246 group-executable mode bits enabled; and
3247 • the caller has permission to read and write the target
3249 file (either via the file's permissions mask or because
3250 it has suitable capabilities).
3251 The default value in this file is 0. Setting the value to 1
3253 prevents a longstanding class of security issues caused by hard-
3254 link-based time-of-check, time-of-use races, most commonly seen
3255 in world-writable directories such as /tmp. The common method
3256 of exploiting this flaw is to cross privilege boundaries when
3257 following a given hard link (i.e., a root process follows a hard
3258 link created by another user). Additionally, on systems without
3259 separated partitions, this stops unauthorized users from "pin‐
3260 ning" vulnerable set-user-ID and set-group-ID files against be‐
3261 ing upgraded by the administrator, or linking to special files.
3262 /proc/sys/fs/protected_regular (since Linux 4.19)
3264 The value in this file is/can be set to one of the following:
3265 0 Writing to regular files is unrestricted.
3267 1 Don't allow O_CREAT open(2) on regular files that the caller
3269 doesn't own in world-writable sticky directories, unless the
3270 regular file is owned by the owner of the directory.
3271 2 As for the value 1, but the restriction also applies to
3273 group-writable sticky directories.
3274 The intent of the above protections is similar to protected_fi‐
3276 fos, but allows an application to avoid writes to an attacker-
3277 controlled regular file, where the application expected to cre‐
3278 ate one.
3279 /proc/sys/fs/protected_symlinks (since Linux 3.6)
3281 When the value in this file is 0, no restrictions are placed on
3282 following symbolic links (i.e., this is the historical behavior
3283 before Linux 3.6). When the value in this file is 1, symbolic
3284 links are followed only in the following circumstances:
3285 • the filesystem UID of the process following the link matches
3287 the owner (UID) of the symbolic link (as described in creden‐
3288 tials(7), a process's filesystem UID is normally the same as
3289 its effective UID);
3290 • the link is not in a sticky world-writable directory; or
3292 • the symbolic link and its parent directory have the same
3294 owner (UID)
3295 A system call that fails to follow a symbolic link because of
3297 the above restrictions returns the error EACCES in errno.
3298 The default value in this file is 0. Setting the value to 1
3300 avoids a longstanding class of security issues based on time-of-
3301 check, time-of-use races when accessing symbolic links.
3302 /proc/sys/fs/suid_dumpable (since Linux 2.6.13)
3304 The value in this file is assigned to a process's "dumpable"
3305 flag in the circumstances described in prctl(2). In effect, the
3306 value in this file determines whether core dump files are pro‐
3307 duced for set-user-ID or otherwise protected/tainted binaries.
3308 The "dumpable" setting also affects the ownership of files in a
3309 process's /proc/pid directory, as described above.
3310 Three different integer values can be specified:
3312 0 (default)
3314 This provides the traditional (pre-Linux 2.6.13) behav‐
3315 ior. A core dump will not be produced for a process
3316 which has changed credentials (by calling seteuid(2),
3317 setgid(2), or similar, or by executing a set-user-ID or
3318 set-group-ID program) or whose binary does not have read
3319 permission enabled.
3320 1 ("debug")
3322 All processes dump core when possible. (Reasons why a
3323 process might nevertheless not dump core are described in
3324 core(5).) The core dump is owned by the filesystem user
3325 ID of the dumping process and no security is applied.
3326 This is intended for system debugging situations only:
3327 this mode is insecure because it allows unprivileged
3328 users to examine the memory contents of privileged pro‐
3329 cesses.
3330 2 ("suidsafe")
3332 Any binary which normally would not be dumped (see "0"
3333 above) is dumped readable by root only. This allows the
3334 user to remove the core dump file but not to read it.
3335 For security reasons core dumps in this mode will not
3336 overwrite one another or other files. This mode is ap‐
3337 propriate when administrators are attempting to debug
3338 problems in a normal environment.
3339 Additionally, since Linux 3.6, /proc/sys/kernel/core_pat‐
3341 tern must either be an absolute pathname or a pipe com‐
3342 mand, as detailed in core(5). Warnings will be written
3343 to the kernel log if core_pattern does not follow these
3344 rules, and no core dump will be produced.
3345 For details of the effect of a process's "dumpable" setting on
3347 ptrace access mode checking, see ptrace(2).
3348 /proc/sys/fs/super-max
3350 This file controls the maximum number of superblocks, and thus
3351 the maximum number of mounted filesystems the kernel can have.
3352 You need increase only super-max if you need to mount more
3353 filesystems than the current value in super-max allows you to.
3354 /proc/sys/fs/super-nr
3356 This file contains the number of filesystems currently mounted.
3357 /proc/sys/kernel
3359 This directory contains files controlling a range of kernel pa‐
3360 rameters, as described below.
3361 /proc/sys/kernel/acct
3363 This file contains three numbers: highwater, lowwater, and fre‐
3364 quency. If BSD-style process accounting is enabled, these val‐
3365 ues control its behavior. If free space on filesystem where the
3366 log lives goes below lowwater percent, accounting suspends. If
3367 free space gets above highwater percent, accounting resumes.
3368 frequency determines how often the kernel checks the amount of
3369 free space (value is in seconds). Default values are 4, 2, and
3370 30. That is, suspend accounting if 2% or less space is free;
3371 resume it if 4% or more space is free; consider information
3372 about amount of free space valid for 30 seconds.
3373 /proc/sys/kernel/auto_msgmni (Linux 2.6.27 to Linux 3.18)
3375 From Linux 2.6.27 to Linux 3.18, this file was used to control
3376 recomputing of the value in /proc/sys/kernel/msgmni upon the ad‐
3377 dition or removal of memory or upon IPC namespace creation/re‐
3378 moval. Echoing "1" into this file enabled msgmni automatic re‐
3379 computing (and triggered a recomputation of msgmni based on the
3380 current amount of available memory and number of IPC name‐
3381 spaces). Echoing "0" disabled automatic recomputing. (Auto‐
3382 matic recomputing was also disabled if a value was explicitly
3383 assigned to /proc/sys/kernel/msgmni.) The default value in
3384 auto_msgmni was 1.
3385 Since Linux 3.19, the content of this file has no effect (be‐
3387 cause msgmni defaults to near the maximum value possible), and
3388 reads from this file always return the value "0".
3389 /proc/sys/kernel/cap_last_cap (since Linux 3.2)
3391 See capabilities(7).
3392 /proc/sys/kernel/cap-bound (from Linux 2.2 to Linux 2.6.24)
3394 This file holds the value of the kernel capability bounding set
3395 (expressed as a signed decimal number). This set is ANDed
3396 against the capabilities permitted to a process during ex‐
3397 ecve(2). Starting with Linux 2.6.25, the system-wide capability
3398 bounding set disappeared, and was replaced by a per-thread
3399 bounding set; see capabilities(7).
3400 /proc/sys/kernel/core_pattern
3402 See core(5).
3403 /proc/sys/kernel/core_pipe_limit
3405 See core(5).
3406 /proc/sys/kernel/core_uses_pid
3408 See core(5).
3409 /proc/sys/kernel/ctrl-alt-del
3411 This file controls the handling of Ctrl-Alt-Del from the key‐
3412 board. When the value in this file is 0, Ctrl-Alt-Del is
3413 trapped and sent to the init(1) program to handle a graceful
3414 restart. When the value is greater than zero, Linux's reaction
3415 to a Vulcan Nerve Pinch (tm) will be an immediate reboot, with‐
3416 out even syncing its dirty buffers. Note: when a program (like
3417 dosemu) has the keyboard in "raw" mode, the ctrl-alt-del is in‐
3418 tercepted by the program before it ever reaches the kernel tty
3419 layer, and it's up to the program to decide what to do with it.
3420 /proc/sys/kernel/dmesg_restrict (since Linux 2.6.37)
3422 The value in this file determines who can see kernel syslog con‐
3423 tents. A value of 0 in this file imposes no restrictions. If
3424 the value is 1, only privileged users can read the kernel sys‐
3425 log. (See syslog(2) for more details.) Since Linux 3.4, only
3426 users with the CAP_SYS_ADMIN capability may change the value in
3427 this file.
3428 /proc/sys/kernel/domainname and /proc/sys/kernel/hostname
3430 can be used to set the NIS/YP domainname and the hostname of
3431 your box in exactly the same way as the commands domainname(1)
3432 and hostname(1), that is:
3433 # echo 'darkstar' > /proc/sys/kernel/hostname
3435 # echo 'mydomain' > /proc/sys/kernel/domainname
3436 has the same effect as
3438 # hostname 'darkstar'
3440 # domainname 'mydomain'
3441 Note, however, that the classic darkstar.frop.org has the host‐
3443 name "darkstar" and DNS (Internet Domain Name Server) domainname
3444 "frop.org", not to be confused with the NIS (Network Information
3445 Service) or YP (Yellow Pages) domainname. These two domain
3446 names are in general different. For a detailed discussion see
3447 the hostname(1) man page.
3448 /proc/sys/kernel/hotplug
3450 This file contains the pathname for the hotplug policy agent.
3451 The default value in this file is /sbin/hotplug.
3452 /proc/sys/kernel/htab-reclaim (before Linux 2.4.9.2)
3454 (PowerPC only) If this file is set to a nonzero value, the Pow‐
3455 erPC htab (see kernel file Documentation/powerpc/ppc_htab.txt)
3456 is pruned each time the system hits the idle loop.
3457 /proc/sys/kernel/keys/*
3459 This directory contains various files that define parameters and
3460 limits for the key-management facility. These files are de‐
3461 scribed in keyrings(7).
3462 /proc/sys/kernel/kptr_restrict (since Linux 2.6.38)
3464 The value in this file determines whether kernel addresses are
3465 exposed via /proc files and other interfaces. A value of 0 in
3466 this file imposes no restrictions. If the value is 1, kernel
3467 pointers printed using the %pK format specifier will be replaced
3468 with zeros unless the user has the CAP_SYSLOG capability. If
3469 the value is 2, kernel pointers printed using the %pK format
3470 specifier will be replaced with zeros regardless of the user's
3471 capabilities. The initial default value for this file was 1,
3472 but the default was changed to 0 in Linux 2.6.39. Since Linux
3473 3.4, only users with the CAP_SYS_ADMIN capability can change the
3474 value in this file.
3475 /proc/sys/kernel/l2cr
3477 (PowerPC only) This file contains a flag that controls the L2
3478 cache of G3 processor boards. If 0, the cache is disabled. En‐
3479 abled if nonzero.
3480 /proc/sys/kernel/modprobe
3482 This file contains the pathname for the kernel module loader.
3483 The default value is /sbin/modprobe. The file is present only
3484 if the kernel is built with the CONFIG_MODULES (CONFIG_KMOD in
3485 Linux 2.6.26 and earlier) option enabled. It is described by
3486 the Linux kernel source file Documentation/kmod.txt (present
3487 only in Linux 2.4 and earlier).
3488 /proc/sys/kernel/modules_disabled (since Linux 2.6.31)
3490 A toggle value indicating if modules are allowed to be loaded in
3491 an otherwise modular kernel. This toggle defaults to off (0),
3492 but can be set true (1). Once true, modules can be neither
3493 loaded nor unloaded, and the toggle cannot be set back to false.
3494 The file is present only if the kernel is built with the CON‐
3495 FIG_MODULES option enabled.
3496 /proc/sys/kernel/msgmax (since Linux 2.2)
3498 This file defines a system-wide limit specifying the maximum
3499 number of bytes in a single message written on a System V mes‐
3500 sage queue.
3501 /proc/sys/kernel/msgmni (since Linux 2.4)
3503 This file defines the system-wide limit on the number of message
3504 queue identifiers. See also /proc/sys/kernel/auto_msgmni.
3505 /proc/sys/kernel/msgmnb (since Linux 2.2)
3507 This file defines a system-wide parameter used to initialize the
3508 msg_qbytes setting for subsequently created message queues. The
3509 msg_qbytes setting specifies the maximum number of bytes that
3510 may be written to the message queue.
3511 /proc/sys/kernel/ngroups_max (since Linux 2.6.4)
3513 This is a read-only file that displays the upper limit on the
3514 number of a process's group memberships.
3515 /proc/sys/kernel/ns_last_pid (since Linux 3.3)
3517 See pid_namespaces(7).
3518 /proc/sys/kernel/ostype and /proc/sys/kernel/osrelease
3520 These files give substrings of /proc/version.
3521 /proc/sys/kernel/overflowgid and /proc/sys/kernel/overflowuid
3523 These files duplicate the files /proc/sys/fs/overflowgid and
3524 /proc/sys/fs/overflowuid.
3525 /proc/sys/kernel/panic
3527 This file gives read/write access to the kernel variable
3528 panic_timeout. If this is zero, the kernel will loop on a
3529 panic; if nonzero, it indicates that the kernel should autore‐
3530 boot after this number of seconds. When you use the software
3531 watchdog device driver, the recommended setting is 60.
3532 /proc/sys/kernel/panic_on_oops (since Linux 2.5.68)
3534 This file controls the kernel's behavior when an oops or BUG is
3535 encountered. If this file contains 0, then the system tries to
3536 continue operation. If it contains 1, then the system delays a
3537 few seconds (to give klogd time to record the oops output) and
3538 then panics. If the /proc/sys/kernel/panic file is also non‐
3539 zero, then the machine will be rebooted.
3540 /proc/sys/kernel/pid_max (since Linux 2.5.34)
3542 This file specifies the value at which PIDs wrap around (i.e.,
3543 the value in this file is one greater than the maximum PID).
3544 PIDs greater than this value are not allocated; thus, the value
3545 in this file also acts as a system-wide limit on the total num‐
3546 ber of processes and threads. The default value for this file,
3547 32768, results in the same range of PIDs as on earlier kernels.
3548 On 32-bit platforms, 32768 is the maximum value for pid_max. On
3549 64-bit systems, pid_max can be set to any value up to 2^22
3550 (PID_MAX_LIMIT, approximately 4 million).
3551 /proc/sys/kernel/powersave-nap (PowerPC only)
3553 This file contains a flag. If set, Linux-PPC will use the "nap"
3554 mode of powersaving, otherwise the "doze" mode will be used.
3555 /proc/sys/kernel/printk
3557 See syslog(2).
3558 /proc/sys/kernel/pty (since Linux 2.6.4)
3560 This directory contains two files relating to the number of UNIX
3561 98 pseudoterminals (see pts(4)) on the system.
3562 /proc/sys/kernel/pty/max
3564 This file defines the maximum number of pseudoterminals.
3565 /proc/sys/kernel/pty/nr
3567 This read-only file indicates how many pseudoterminals are cur‐
3568 rently in use.
3569 /proc/sys/kernel/random
3571 This directory contains various parameters controlling the oper‐
3572 ation of the file /dev/random. See random(4) for further infor‐
3573 mation.
3574 /proc/sys/kernel/random/uuid (since Linux 2.4)
3576 Each read from this read-only file returns a randomly generated
3577 128-bit UUID, as a string in the standard UUID format.
3578 /proc/sys/kernel/randomize_va_space (since Linux 2.6.12)
3580 Select the address space layout randomization (ASLR) policy for
3581 the system (on architectures that support ASLR). Three values
3582 are supported for this file:
3583 0 Turn ASLR off. This is the default for architectures
3585 that don't support ASLR, and when the kernel is booted
3586 with the norandmaps parameter.
3587 1 Make the addresses of mmap(2) allocations, the stack, and
3589 the VDSO page randomized. Among other things, this means
3590 that shared libraries will be loaded at randomized ad‐
3591 dresses. The text segment of PIE-linked binaries will
3592 also be loaded at a randomized address. This value is
3593 the default if the kernel was configured with CONFIG_COM‐
3594 PAT_BRK.
3595 2 (Since Linux 2.6.25) Also support heap randomization.
3597 This value is the default if the kernel was not config‐
3598 ured with CONFIG_COMPAT_BRK.
3599 /proc/sys/kernel/real-root-dev
3601 This file is documented in the Linux kernel source file Documen‐
3602 tation/admin-guide/initrd.rst (or Documentation/initrd.txt be‐
3603 fore Linux 4.10).
3604 /proc/sys/kernel/reboot-cmd (Sparc only)
3606 This file seems to be a way to give an argument to the SPARC
3607 ROM/Flash boot loader. Maybe to tell it what to do after re‐
3608 booting?
3609 /proc/sys/kernel/rtsig-max
3611 (Up to and including Linux 2.6.7; see setrlimit(2)) This file
3612 can be used to tune the maximum number of POSIX real-time
3613 (queued) signals that can be outstanding in the system.
3614 /proc/sys/kernel/rtsig-nr
3616 (Up to and including Linux 2.6.7.) This file shows the number
3617 of POSIX real-time signals currently queued.
3618 /proc/pid/sched_autogroup_enabled (since Linux 2.6.38)
3620 See sched(7).
3621 /proc/sys/kernel/sched_child_runs_first (since Linux 2.6.23)
3623 If this file contains the value zero, then, after a fork(2), the
3624 parent is first scheduled on the CPU. If the file contains a
3625 nonzero value, then the child is scheduled first on the CPU.
3626 (Of course, on a multiprocessor system, the parent and the child
3627 might both immediately be scheduled on a CPU.)
3628 /proc/sys/kernel/sched_rr_timeslice_ms (since Linux 3.9)
3630 See sched_rr_get_interval(2).
3631 /proc/sys/kernel/sched_rt_period_us (since Linux 2.6.25)
3633 See sched(7).
3634 /proc/sys/kernel/sched_rt_runtime_us (since Linux 2.6.25)
3636 See sched(7).
3637 /proc/sys/kernel/seccomp (since Linux 4.14)
3639 This directory provides additional seccomp information and con‐
3640 figuration. See seccomp(2) for further details.
3641 /proc/sys/kernel/sem (since Linux 2.4)
3643 This file contains 4 numbers defining limits for System V IPC
3644 semaphores. These fields are, in order:
3645 SEMMSL The maximum semaphores per semaphore set.
3647 SEMMNS A system-wide limit on the number of semaphores in all
3649 semaphore sets.
3650 SEMOPM The maximum number of operations that may be specified in
3652 a semop(2) call.
3653 SEMMNI A system-wide limit on the maximum number of semaphore
3655 identifiers.
3656 /proc/sys/kernel/sg-big-buff
3658 This file shows the size of the generic SCSI device (sg) buffer.
3659 You can't tune it just yet, but you could change it at compile
3660 time by editing include/scsi/sg.h and changing the value of
3661 SG_BIG_BUFF. However, there shouldn't be any reason to change
3662 this value.
3663 /proc/sys/kernel/shm_rmid_forced (since Linux 3.1)
3665 If this file is set to 1, all System V shared memory segments
3666 will be marked for destruction as soon as the number of attached
3667 processes falls to zero; in other words, it is no longer possi‐
3668 ble to create shared memory segments that exist independently of
3669 any attached process.
3670 The effect is as though a shmctl(2) IPC_RMID is performed on all
3672 existing segments as well as all segments created in the future
3673 (until this file is reset to 0). Note that existing segments
3674 that are attached to no process will be immediately destroyed
3675 when this file is set to 1. Setting this option will also de‐
3676 stroy segments that were created, but never attached, upon ter‐
3677 mination of the process that created the segment with shmget(2).
3678 Setting this file to 1 provides a way of ensuring that all Sys‐
3680 tem V shared memory segments are counted against the resource
3681 usage and resource limits (see the description of RLIMIT_AS in
3682 getrlimit(2)) of at least one process.
3683 Because setting this file to 1 produces behavior that is non‐
3685 standard and could also break existing applications, the default
3686 value in this file is 0. Set this file to 1 only if you have a
3687 good understanding of the semantics of the applications using
3688 System V shared memory on your system.
3689 /proc/sys/kernel/shmall (since Linux 2.2)
3691 This file contains the system-wide limit on the total number of
3692 pages of System V shared memory.
3693 /proc/sys/kernel/shmmax (since Linux 2.2)
3695 This file can be used to query and set the run-time limit on the
3696 maximum (System V IPC) shared memory segment size that can be
3697 created. Shared memory segments up to 1 GB are now supported in
3698 the kernel. This value defaults to SHMMAX.
3699 /proc/sys/kernel/shmmni (since Linux 2.4)
3701 This file specifies the system-wide maximum number of System V
3702 shared memory segments that can be created.
3703 /proc/sys/kernel/sysctl_writes_strict (since Linux 3.16)
3705 The value in this file determines how the file offset affects
3706 the behavior of updating entries in files under /proc/sys. The
3707 file has three possible values:
3708 -1 This provides legacy handling, with no printk warnings.
3710 Each write(2) must fully contain the value to be written,
3711 and multiple writes on the same file descriptor will over‐
3712 write the entire value, regardless of the file position.
3713 0 (default) This provides the same behavior as for -1, but
3715 printk warnings are written for processes that perform
3716 writes when the file offset is not 0.
3717 1 Respect the file offset when writing strings into /proc/sys
3719 files. Multiple writes will append to the value buffer.
3720 Anything written beyond the maximum length of the value buf‐
3721 fer will be ignored. Writes to numeric /proc/sys entries
3722 must always be at file offset 0 and the value must be fully
3723 contained in the buffer provided to write(2).
3724 /proc/sys/kernel/sysrq
3726 This file controls the functions allowed to be invoked by the
3727 SysRq key. By default, the file contains 1 meaning that every
3728 possible SysRq request is allowed (in older kernel versions,
3729 SysRq was disabled by default, and you were required to specifi‐
3730 cally enable it at run-time, but this is not the case any more).
3731 Possible values in this file are:
3732 0 Disable sysrq completely
3734 1 Enable all functions of sysrq
3736 > 1 Bit mask of allowed sysrq functions, as follows:
3738 2 Enable control of console logging level
3739 4 Enable control of keyboard (SAK, unraw)
3740 8 Enable debugging dumps of processes etc.
3741 16 Enable sync command
3742 32 Enable remount read-only
3743 64 Enable signaling of processes (term, kill, oom-kill)
3744 128 Allow reboot/poweroff
3745 256 Allow nicing of all real-time tasks
3746 This file is present only if the CONFIG_MAGIC_SYSRQ kernel con‐
3748 figuration option is enabled. For further details see the Linux
3749 kernel source file Documentation/admin-guide/sysrq.rst (or Docu‐
3750 mentation/sysrq.txt before Linux 4.10).
3751 /proc/sys/kernel/version
3753 This file contains a string such as:
3754 #5 Wed Feb 25 21:49:24 MET 1998
3756 The "#5" means that this is the fifth kernel built from this
3758 source base and the date following it indicates the time the
3759 kernel was built.
3760 /proc/sys/kernel/threads-max (since Linux 2.3.11)
3762 This file specifies the system-wide limit on the number of
3763 threads (tasks) that can be created on the system.
3764 Since Linux 4.1, the value that can be written to threads-max is
3766 bounded. The minimum value that can be written is 20. The max‐
3767 imum value that can be written is given by the constant FU‐
3768 TEX_TID_MASK [22m(0x3fffffff). If a value outside of this range is
3769 written to threads-max, the error EINVAL occurs.
3770 The value written is checked against the available RAM pages.
3772 If the thread structures would occupy too much (more than 1/8th)
3773 of the available RAM pages, threads-max is reduced accordingly.
3774 /proc/sys/kernel/yama/ptrace_scope (since Linux 3.5)
3776 See ptrace(2).
3777 /proc/sys/kernel/zero-paged (PowerPC only)
3779 This file contains a flag. When enabled (nonzero), Linux-PPC
3780 will pre-zero pages in the idle loop, possibly speeding up
3781 get_free_pages.
3782 /proc/sys/net
3784 This directory contains networking stuff. Explanations for some
3785 of the files under this directory can be found in tcp(7) and
3786 ip(7).
3787 /proc/sys/net/core/bpf_jit_enable
3789 See bpf(2).
3790 /proc/sys/net/core/somaxconn
3792 This file defines a ceiling value for the backlog argument of
3793 listen(2); see the listen(2) manual page for details.
3794 /proc/sys/proc
3796 This directory may be empty.
3797 /proc/sys/sunrpc
3799 This directory supports Sun remote procedure call for network
3800 filesystem (NFS). On some systems, it is not present.
3801 /proc/sys/user (since Linux 4.9)
3803 See namespaces(7).
3804 /proc/sys/vm
3806 This directory contains files for memory management tuning, buf‐
3807 fer, and cache management.
3808 /proc/sys/vm/admin_reserve_kbytes (since Linux 3.10)
3810 This file defines the amount of free memory (in KiB) on the sys‐
3811 tem that should be reserved for users with the capability
3812 CAP_SYS_ADMIN.
3813 The default value in this file is the minimum of [3% of free
3815 pages, 8MiB] expressed as KiB. The default is intended to pro‐
3816 vide enough for the superuser to log in and kill a process, if
3817 necessary, under the default overcommit 'guess' mode (i.e., 0 in
3818 /proc/sys/vm/overcommit_memory).
3819 Systems running in "overcommit never" mode (i.e., 2 in
3821 /proc/sys/vm/overcommit_memory) should increase the value in
3822 this file to account for the full virtual memory size of the
3823 programs used to recover (e.g., login(1) ssh(1), and top(1))
3824 Otherwise, the superuser may not be able to log in to recover
3825 the system. For example, on x86-64 a suitable value is 131072
3826 (128MiB reserved).
3827 Changing the value in this file takes effect whenever an appli‐
3829 cation requests memory.
3830 /proc/sys/vm/compact_memory (since Linux 2.6.35)
3832 When 1 is written to this file, all zones are compacted such
3833 that free memory is available in contiguous blocks where possi‐
3834 ble. The effect of this action can be seen by examining
3835 /proc/buddyinfo.
3836 Present only if the kernel was configured with CONFIG_COM‐
3838 PACTION.
3839 /proc/sys/vm/drop_caches (since Linux 2.6.16)
3841 Writing to this file causes the kernel to drop clean caches,
3842 dentries, and inodes from memory, causing that memory to become
3843 free. This can be useful for memory management testing and per‐
3844 forming reproducible filesystem benchmarks. Because writing to
3845 this file causes the benefits of caching to be lost, it can de‐
3846 grade overall system performance.
3847 To free pagecache, use:
3849 echo 1 > /proc/sys/vm/drop_caches
3851 To free dentries and inodes, use:
3853 echo 2 > /proc/sys/vm/drop_caches
3855 To free pagecache, dentries, and inodes, use:
3857 echo 3 > /proc/sys/vm/drop_caches
3859 Because writing to this file is a nondestructive operation and
3861 dirty objects are not freeable, the user should run sync(1)
3862 first.
3863 /proc/sys/vm/sysctl_hugetlb_shm_group (since Linux 2.6.7)
3865 This writable file contains a group ID that is allowed to allo‐
3866 cate memory using huge pages. If a process has a filesystem
3867 group ID or any supplementary group ID that matches this group
3868 ID, then it can make huge-page allocations without holding the
3869 CAP_IPC_LOCK capability; see memfd_create(2), mmap(2), and
3870 shmget(2).
3871 /proc/sys/vm/legacy_va_layout (since Linux 2.6.9)
3873 If nonzero, this disables the new 32-bit memory-mapping layout;
3874 the kernel will use the legacy (2.4) layout for all processes.
3875 /proc/sys/vm/memory_failure_early_kill (since Linux 2.6.32)
3877 Control how to kill processes when an uncorrected memory error
3878 (typically a 2-bit error in a memory module) that cannot be han‐
3879 dled by the kernel is detected in the background by hardware.
3880 In some cases (like the page still having a valid copy on disk),
3881 the kernel will handle the failure transparently without affect‐
3882 ing any applications. But if there is no other up-to-date copy
3883 of the data, it will kill processes to prevent any data corrup‐
3884 tions from propagating.
3885 The file has one of the following values:
3887 1 Kill all processes that have the corrupted-and-not-
3889 reloadable page mapped as soon as the corruption is de‐
3890 tected. Note that this is not supported for a few types
3891 of pages, such as kernel internally allocated data or the
3892 swap cache, but works for the majority of user pages.
3893 0 Unmap the corrupted page from all processes and kill a
3895 process only if it tries to access the page.
3896 The kill is performed using a SIGBUS signal with si_code set to
3898 BUS_MCEERR_AO. Processes can handle this if they want to; see
3899 sigaction(2) for more details.
3900 This feature is active only on architectures/platforms with ad‐
3902 vanced machine check handling and depends on the hardware capa‐
3903 bilities.
3904 Applications can override the memory_failure_early_kill setting
3906 individually with the prctl(2) PR_MCE_KILL operation.
3907 Present only if the kernel was configured with CONFIG_MEM‐
3909 ORY_FAILURE.
3910 /proc/sys/vm/memory_failure_recovery (since Linux 2.6.32)
3912 Enable memory failure recovery (when supported by the platform).
3913 1 Attempt recovery.
3915 0 Always panic on a memory failure.
3917 Present only if the kernel was configured with CONFIG_MEM‐
3919 ORY_FAILURE.
3920 /proc/sys/vm/oom_dump_tasks (since Linux 2.6.25)
3922 Enables a system-wide task dump (excluding kernel threads) to be
3923 produced when the kernel performs an OOM-killing. The dump in‐
3924 cludes the following information for each task (thread,
3925 process): thread ID, real user ID, thread group ID (process ID),
3926 virtual memory size, resident set size, the CPU that the task is
3927 scheduled on, oom_adj score (see the description of
3928 /proc/pid/oom_adj), and command name. This is helpful to deter‐
3929 mine why the OOM-killer was invoked and to identify the rogue
3930 task that caused it.
3931 If this contains the value zero, this information is suppressed.
3933 On very large systems with thousands of tasks, it may not be
3934 feasible to dump the memory state information for each one.
3935 Such systems should not be forced to incur a performance penalty
3936 in OOM situations when the information may not be desired.
3937 If this is set to nonzero, this information is shown whenever
3939 the OOM-killer actually kills a memory-hogging task.
3940 The default value is 0.
3942 /proc/sys/vm/oom_kill_allocating_task (since Linux 2.6.24)
3944 This enables or disables killing the OOM-triggering task in out-
3945 of-memory situations.
3946 If this is set to zero, the OOM-killer will scan through the en‐
3948 tire tasklist and select a task based on heuristics to kill.
3949 This normally selects a rogue memory-hogging task that frees up
3950 a large amount of memory when killed.
3951 If this is set to nonzero, the OOM-killer simply kills the task
3953 that triggered the out-of-memory condition. This avoids a pos‐
3954 sibly expensive tasklist scan.
3955 If /proc/sys/vm/panic_on_oom is nonzero, it takes precedence
3957 over whatever value is used in /proc/sys/vm/oom_kill_allocat‐
3958 ing_task.
3959 The default value is 0.
3961 /proc/sys/vm/overcommit_kbytes (since Linux 3.14)
3963 This writable file provides an alternative to /proc/sys/vm/over‐
3964 commit_ratio for controlling the CommitLimit when
3965 /proc/sys/vm/overcommit_memory has the value 2. It allows the
3966 amount of memory overcommitting to be specified as an absolute
3967 value (in kB), rather than as a percentage, as is done with
3968 overcommit_ratio. This allows for finer-grained control of Com‐
3969 mitLimit on systems with extremely large memory sizes.
3970 Only one of overcommit_kbytes or overcommit_ratio can have an
3972 effect: if overcommit_kbytes has a nonzero value, then it is
3973 used to calculate CommitLimit, otherwise overcommit_ratio is
3974 used. Writing a value to either of these files causes the value
3975 in the other file to be set to zero.
3976 /proc/sys/vm/overcommit_memory
3978 This file contains the kernel virtual memory accounting mode.
3979 Values are:
3980 0: heuristic overcommit (this is the default)
3982 1: always overcommit, never check
3983 2: always check, never overcommit
3984 In mode 0, calls of mmap(2) with MAP_NORESERVE are not checked,
3986 and the default check is very weak, leading to the risk of get‐
3987 ting a process "OOM-killed".
3988 In mode 1, the kernel pretends there is always enough memory,
3990 until memory actually runs out. One use case for this mode is
3991 scientific computing applications that employ large sparse ar‐
3992 rays. Before Linux 2.6.0, any nonzero value implies mode 1.
3993 In mode 2 (available since Linux 2.6), the total virtual address
3995 space that can be allocated (CommitLimit in /proc/meminfo) is
3996 calculated as
3997 CommitLimit = (total_RAM - total_huge_TLB) *
3999 overcommit_ratio / 100 + total_swap
4000 where:
4002 • total_RAM is the total amount of RAM on the system;
4004 • total_huge_TLB is the amount of memory set aside for huge
4006 pages;
4007 • overcommit_ratio is the value in /proc/sys/vm/overcommit_ra‐
4009 tio; and
4010 • total_swap is the amount of swap space.
4012 For example, on a system with 16 GB of physical RAM, 16 GB of
4014 swap, no space dedicated to huge pages, and an overcommit_ratio
4015 of 50, this formula yields a CommitLimit of 24 GB.
4016 Since Linux 3.14, if the value in /proc/sys/vm/overcommit_kbytes
4018 is nonzero, then CommitLimit is instead calculated as:
4019 CommitLimit = overcommit_kbytes + total_swap
4021 See also the description of /proc/sys/vm/admin_reserve_kbytes
4023 and /proc/sys/vm/user_reserve_kbytes.
4024 /proc/sys/vm/overcommit_ratio (since Linux 2.6.0)
4026 This writable file defines a percentage by which memory can be
4027 overcommitted. The default value in the file is 50. See the
4028 description of /proc/sys/vm/overcommit_memory.
4029 /proc/sys/vm/panic_on_oom (since Linux 2.6.18)
4031 This enables or disables a kernel panic in an out-of-memory sit‐
4032 uation.
4033 If this file is set to the value 0, the kernel's OOM-killer will
4035 kill some rogue process. Usually, the OOM-killer is able to
4036 kill a rogue process and the system will survive.
4037 If this file is set to the value 1, then the kernel normally
4039 panics when out-of-memory happens. However, if a process limits
4040 allocations to certain nodes using memory policies (mbind(2)
4041 MPOL_BIND) or cpusets (cpuset(7)) and those nodes reach memory
4042 exhaustion status, one process may be killed by the OOM-killer.
4043 No panic occurs in this case: because other nodes' memory may be
4044 free, this means the system as a whole may not have reached an
4045 out-of-memory situation yet.
4046 If this file is set to the value 2, the kernel always panics
4048 when an out-of-memory condition occurs.
4049 The default value is 0. 1 and 2 are for failover of clustering.
4051 Select either according to your policy of failover.
4052 /proc/sys/vm/swappiness
4054 The value in this file controls how aggressively the kernel will
4055 swap memory pages. Higher values increase aggressiveness, lower
4056 values decrease aggressiveness. The default value is 60.
4057 /proc/sys/vm/user_reserve_kbytes (since Linux 3.10)
4059 Specifies an amount of memory (in KiB) to reserve for user pro‐
4060 cesses. This is intended to prevent a user from starting a sin‐
4061 gle memory hogging process, such that they cannot recover (kill
4062 the hog). The value in this file has an effect only when
4063 /proc/sys/vm/overcommit_memory is set to 2 ("overcommit never"
4064 mode). In this case, the system reserves an amount of memory
4065 that is the minimum of [3% of current process size, user_re‐
4066 serve_kbytes].
4067 The default value in this file is the minimum of [3% of free
4069 pages, 128MiB] expressed as KiB.
4070 If the value in this file is set to zero, then a user will be
4072 allowed to allocate all free memory with a single process (minus
4073 the amount reserved by /proc/sys/vm/admin_reserve_kbytes). Any
4074 subsequent attempts to execute a command will result in "fork:
4075 Cannot allocate memory".
4076 Changing the value in this file takes effect whenever an appli‐
4078 cation requests memory.
4079 /proc/sys/vm/unprivileged_userfaultfd (since Linux 5.2)
4081 This (writable) file exposes a flag that controls whether un‐
4082 privileged processes are allowed to employ userfaultfd(2). If
4083 this file has the value 1, then unprivileged processes may use
4084 userfaultfd(2). If this file has the value 0, then only pro‐
4085 cesses that have the CAP_SYS_PTRACE capability may employ user‐
4086 faultfd(2). The default value in this file is 1.
4087 /proc/sysrq-trigger (since Linux 2.4.21)
4089 Writing a character to this file triggers the same SysRq func‐
4090 tion as typing ALT-SysRq-<character> (see the description of
4091 /proc/sys/kernel/sysrq). This file is normally writable only by
4092 root. For further details see the Linux kernel source file Doc‐
4093 umentation/admin-guide/sysrq.rst (or Documentation/sysrq.txt be‐
4094 fore Linux 4.10).
4095 /proc/sysvipc
4097 Subdirectory containing the pseudo-files msg, sem and shm.
4098 These files list the System V Interprocess Communication (IPC)
4099 objects (respectively: message queues, semaphores, and shared
4100 memory) that currently exist on the system, providing similar
4101 information to that available via ipcs(1). These files have
4102 headers and are formatted (one IPC object per line) for easy un‐
4103 derstanding. sysvipc(7) provides further background on the in‐
4104 formation shown by these files.
4105 /proc/thread-self (since Linux 3.17)
4107 This directory refers to the thread accessing the /proc filesys‐
4108 tem, and is identical to the /proc/self/task/tid directory named
4109 by the process thread ID (tid) of the same thread.
4110 /proc/timer_list (since Linux 2.6.21)
4112 This read-only file exposes a list of all currently pending
4113 (high-resolution) timers, all clock-event sources, and their pa‐
4114 rameters in a human-readable form.
4115 /proc/timer_stats (from Linux 2.6.21 until Linux 4.10)
4117 This is a debugging facility to make timer (ab)use in a Linux
4118 system visible to kernel and user-space developers. It can be
4119 used by kernel and user-space developers to verify that their
4120 code does not make undue use of timers. The goal is to avoid
4121 unnecessary wakeups, thereby optimizing power consumption.
4122 If enabled in the kernel (CONFIG_TIMER_STATS), but not used, it
4124 has almost zero run-time overhead and a relatively small data-
4125 structure overhead. Even if collection is enabled at run time,
4126 overhead is low: all the locking is per-CPU and lookup is
4127 hashed.
4128 The /proc/timer_stats file is used both to control sampling fa‐
4130 cility and to read out the sampled information.
4131 The timer_stats functionality is inactive on bootup. A sampling
4133 period can be started using the following command:
4134 # echo 1 > /proc/timer_stats
4136 The following command stops a sampling period:
4138 # echo 0 > /proc/timer_stats
4140 The statistics can be retrieved by:
4142 $ cat /proc/timer_stats
4144 While sampling is enabled, each readout from /proc/timer_stats
4146 will see newly updated statistics. Once sampling is disabled,
4147 the sampled information is kept until a new sample period is
4148 started. This allows multiple readouts.
4149 Sample output from /proc/timer_stats:
4151 $ cat /proc/timer_stats
4153 Timer Stats Version: v0.3
4154 Sample period: 1.764 s
4155 Collection: active
4156 255, 0 swapper/3 hrtimer_start_range_ns (tick_sched_timer)
4157 71, 0 swapper/1 hrtimer_start_range_ns (tick_sched_timer)
4158 58, 0 swapper/0 hrtimer_start_range_ns (tick_sched_timer)
4159 4, 1694 gnome-shell mod_delayed_work_on (delayed_work_timer_fn)
4160 17, 7 rcu_sched rcu_gp_kthread (process_timeout)
4161 ...
4162 1, 4911 kworker/u16:0 mod_delayed_work_on (delayed_work_timer_fn)
4163 1D, 2522 kworker/0:0 queue_delayed_work_on (delayed_work_timer_fn)
4164 1029 total events, 583.333 events/sec
4165 The output columns are:
4167 [1] a count of the number of events, optionally (since Linux
4169 2.6.23) followed by the letter 'D' if this is a deferrable
4170 timer;
4171 [2] the PID of the process that initialized the timer;
4173 [3] the name of the process that initialized the timer;
4175 [4] the function where the timer was initialized; and (in
4177 parentheses) the callback function that is associated with
4178 the timer.
4179 During the Linux 4.11 development cycle, this file was removed
4181 because of security concerns, as it exposes information across
4182 namespaces. Furthermore, it is possible to obtain the same in‐
4183 formation via in-kernel tracing facilities such as ftrace.
4184 /proc/tty
4186 Subdirectory containing the pseudo-files and subdirectories for
4187 tty drivers and line disciplines.
4188 /proc/uptime
4190 This file contains two numbers (values in seconds): the uptime
4191 of the system (including time spent in suspend) and the amount
4192 of time spent in the idle process.
4193 /proc/version
4195 This string identifies the kernel version that is currently run‐
4196 ning. It includes the contents of /proc/sys/kernel/ostype,
4197 /proc/sys/kernel/osrelease, and /proc/sys/kernel/version. For
4198 example:
4199 Linux version 1.0.9 (quinlan@phaze) #1 Sat May 14 01:51:54 EDT 1994
4201 /proc/vmstat (since Linux 2.6.0)
4203 This file displays various virtual memory statistics. Each line
4204 of this file contains a single name-value pair, delimited by
4205 white space. Some lines are present only if the kernel was con‐
4206 figured with suitable options. (In some cases, the options re‐
4207 quired for particular files have changed across kernel versions,
4208 so they are not listed here. Details can be found by consulting
4209 the kernel source code.) The following fields may be present:
4210 nr_free_pages (since Linux 2.6.31)
4212 nr_alloc_batch (since Linux 3.12)
4214 nr_inactive_anon (since Linux 2.6.28)
4216 nr_active_anon (since Linux 2.6.28)
4218 nr_inactive_file (since Linux 2.6.28)
4220 nr_active_file (since Linux 2.6.28)
4222 nr_unevictable (since Linux 2.6.28)
4224 nr_mlock (since Linux 2.6.28)
4226 nr_anon_pages (since Linux 2.6.18)
4228 nr_mapped (since Linux 2.6.0)
4230 nr_file_pages (since Linux 2.6.18)
4232 nr_dirty (since Linux 2.6.0)
4234 nr_writeback (since Linux 2.6.0)
4236 nr_slab_reclaimable (since Linux 2.6.19)
4238 nr_slab_unreclaimable (since Linux 2.6.19)
4240 nr_page_table_pages (since Linux 2.6.0)
4242 nr_kernel_stack (since Linux 2.6.32)
4244 Amount of memory allocated to kernel stacks.
4245 nr_unstable (since Linux 2.6.0)
4247 nr_bounce (since Linux 2.6.12)
4249 nr_vmscan_write (since Linux 2.6.19)
4251 nr_vmscan_immediate_reclaim (since Linux 3.2)
4253 nr_writeback_temp (since Linux 2.6.26)
4255 nr_isolated_anon (since Linux 2.6.32)
4257 nr_isolated_file (since Linux 2.6.32)
4259 nr_shmem (since Linux 2.6.32)
4261 Pages used by shmem and tmpfs(5).
4262 nr_dirtied (since Linux 2.6.37)
4264 nr_written (since Linux 2.6.37)
4266 nr_pages_scanned (since Linux 3.17)
4268 numa_hit (since Linux 2.6.18)
4270 numa_miss (since Linux 2.6.18)
4272 numa_foreign (since Linux 2.6.18)
4274 numa_interleave (since Linux 2.6.18)
4276 numa_local (since Linux 2.6.18)
4278 numa_other (since Linux 2.6.18)
4280 workingset_refault (since Linux 3.15)
4282 workingset_activate (since Linux 3.15)
4284 workingset_nodereclaim (since Linux 3.15)
4286 nr_anon_transparent_hugepages (since Linux 2.6.38)
4288 nr_free_cma (since Linux 3.7)
4290 Number of free CMA (Contiguous Memory Allocator) pages.
4291 nr_dirty_threshold (since Linux 2.6.37)
4293 nr_dirty_background_threshold (since Linux 2.6.37)
4295 pgpgin (since Linux 2.6.0)
4297 pgpgout (since Linux 2.6.0)
4299 pswpin (since Linux 2.6.0)
4301 pswpout (since Linux 2.6.0)
4303 pgalloc_dma (since Linux 2.6.5)
4305 pgalloc_dma32 (since Linux 2.6.16)
4307 pgalloc_normal (since Linux 2.6.5)
4309 pgalloc_high (since Linux 2.6.5)
4311 pgalloc_movable (since Linux 2.6.23)
4313 pgfree (since Linux 2.6.0)
4315 pgactivate (since Linux 2.6.0)
4317 pgdeactivate (since Linux 2.6.0)
4319 pgfault (since Linux 2.6.0)
4321 pgmajfault (since Linux 2.6.0)
4323 pgrefill_dma (since Linux 2.6.5)
4325 pgrefill_dma32 (since Linux 2.6.16)
4327 pgrefill_normal (since Linux 2.6.5)
4329 pgrefill_high (since Linux 2.6.5)
4331 pgrefill_movable (since Linux 2.6.23)
4333 pgsteal_kswapd_dma (since Linux 3.4)
4335 pgsteal_kswapd_dma32 (since Linux 3.4)
4337 pgsteal_kswapd_normal (since Linux 3.4)
4339 pgsteal_kswapd_high (since Linux 3.4)
4341 pgsteal_kswapd_movable (since Linux 3.4)
4343 pgsteal_direct_dma
4345 pgsteal_direct_dma32 (since Linux 3.4)
4347 pgsteal_direct_normal (since Linux 3.4)
4349 pgsteal_direct_high (since Linux 3.4)
4351 pgsteal_direct_movable (since Linux 2.6.23)
4353 pgscan_kswapd_dma
4355 pgscan_kswapd_dma32 (since Linux 2.6.16)
4357 pgscan_kswapd_normal (since Linux 2.6.5)
4359 pgscan_kswapd_high
4361 pgscan_kswapd_movable (since Linux 2.6.23)
4363 pgscan_direct_dma
4365 pgscan_direct_dma32 (since Linux 2.6.16)
4367 pgscan_direct_normal
4369 pgscan_direct_high
4371 pgscan_direct_movable (since Linux 2.6.23)
4373 pgscan_direct_throttle (since Linux 3.6)
4375 zone_reclaim_failed (since linux 2.6.31)
4377 pginodesteal (since linux 2.6.0)
4379 slabs_scanned (since linux 2.6.5)
4381 kswapd_inodesteal (since linux 2.6.0)
4383 kswapd_low_wmark_hit_quickly (since Linux 2.6.33)
4385 kswapd_high_wmark_hit_quickly (since Linux 2.6.33)
4387 pageoutrun (since Linux 2.6.0)
4389 allocstall (since Linux 2.6.0)
4391 pgrotated (since Linux 2.6.0)
4393 drop_pagecache (since Linux 3.15)
4395 drop_slab (since Linux 3.15)
4397 numa_pte_updates (since Linux 3.8)
4399 numa_huge_pte_updates (since Linux 3.13)
4401 numa_hint_faults (since Linux 3.8)
4403 numa_hint_faults_local (since Linux 3.8)
4405 numa_pages_migrated (since Linux 3.8)
4407 pgmigrate_success (since Linux 3.8)
4409 pgmigrate_fail (since Linux 3.8)
4411 compact_migrate_scanned (since Linux 3.8)
4413 compact_free_scanned (since Linux 3.8)
4415 compact_isolated (since Linux 3.8)
4417 compact_stall (since Linux 2.6.35)
4419 See the kernel source file Documentation/ad‐
4420 min-guide/mm/transhuge.rst.
4421 compact_fail (since Linux 2.6.35)
4423 See the kernel source file Documentation/ad‐
4424 min-guide/mm/transhuge.rst.
4425 compact_success (since Linux 2.6.35)
4427 See the kernel source file Documentation/ad‐
4428 min-guide/mm/transhuge.rst.
4429 htlb_buddy_alloc_success (since Linux 2.6.26)
4431 htlb_buddy_alloc_fail (since Linux 2.6.26)
4433 unevictable_pgs_culled (since Linux 2.6.28)
4435 unevictable_pgs_scanned (since Linux 2.6.28)
4437 unevictable_pgs_rescued (since Linux 2.6.28)
4439 unevictable_pgs_mlocked (since Linux 2.6.28)
4441 unevictable_pgs_munlocked (since Linux 2.6.28)
4443 unevictable_pgs_cleared (since Linux 2.6.28)
4445 unevictable_pgs_stranded (since Linux 2.6.28)
4447 thp_fault_alloc (since Linux 2.6.39)
4449 See the kernel source file Documentation/ad‐
4450 min-guide/mm/transhuge.rst.
4451 thp_fault_fallback (since Linux 2.6.39)
4453 See the kernel source file Documentation/ad‐
4454 min-guide/mm/transhuge.rst.
4455 thp_collapse_alloc (since Linux 2.6.39)
4457 See the kernel source file Documentation/ad‐
4458 min-guide/mm/transhuge.rst.
4459 thp_collapse_alloc_failed (since Linux 2.6.39)
4461 See the kernel source file Documentation/ad‐
4462 min-guide/mm/transhuge.rst.
4463 thp_split (since Linux 2.6.39)
4465 See the kernel source file Documentation/ad‐
4466 min-guide/mm/transhuge.rst.
4467 thp_zero_page_alloc (since Linux 3.8)
4469 See the kernel source file Documentation/ad‐
4470 min-guide/mm/transhuge.rst.
4471 thp_zero_page_alloc_failed (since Linux 3.8)
4473 See the kernel source file Documentation/ad‐
4474 min-guide/mm/transhuge.rst.
4475 balloon_inflate (since Linux 3.18)
4477 balloon_deflate (since Linux 3.18)
4479 balloon_migrate (since Linux 3.18)
4481 nr_tlb_remote_flush (since Linux 3.12)
4483 nr_tlb_remote_flush_received (since Linux 3.12)
4485 nr_tlb_local_flush_all (since Linux 3.12)
4487 nr_tlb_local_flush_one (since Linux 3.12)
4489 vmacache_find_calls (since Linux 3.16)
4491 vmacache_find_hits (since Linux 3.16)
4493 vmacache_full_flushes (since Linux 3.19)
4495 /proc/zoneinfo (since Linux 2.6.13)
4497 This file displays information about memory zones. This is use‐
4498 ful for analyzing virtual memory behavior.
4499
4501 Many files contain strings (e.g., the environment and command line)
4502 that are in the internal format, with subfields terminated by null
4503 bytes ('\0'). When inspecting such files, you may find that the re‐
4504 sults are more readable if you use a command of the following form to
4505 display them:
4506 $ cat file | tr '\000' '\n'
4508 This manual page is incomplete, possibly inaccurate, and is the kind of
4510 thing that needs to be updated very often.
4511
4513 cat(1), dmesg(1), find(1), free(1), htop(1), init(1), ps(1), pstree(1),
4514 tr(1), uptime(1), chroot(2), mmap(2), readlink(2), syslog(2),
4515 slabinfo(5), sysfs(5), hier(7), namespaces(7), time(7), arp(8), hd‐
4516 parm(8), ifconfig(8), lsmod(8), lspci(8), mount(8), netstat(8),
4517 procinfo(8), route(8), sysctl(8)
4518 The Linux kernel source files: Documentation/filesystems/proc.rst, Doc‐
4520 umentation/admin-guide/sysctl/fs.rst, Documentation/ad‐
4521 min-guide/sysctl/kernel.rst, Documentation/admin-guide/sysctl/net.rst,
4522 and Documentation/admin-guide/sysctl/vm.rst.
4523Linux man-pages 6.04 2023-04-02 proc(5)