blob: 66cad5c86171f1bd6b5dffb2878d618dc2d1fbbf [file] [log] [blame]
Linus Torvalds1da177e2005-04-16 15:20:36 -07001------------------------------------------------------------------------------
2 T H E /proc F I L E S Y S T E M
3------------------------------------------------------------------------------
4/proc/sys Terrehon Bowden <terrehon@pacbell.net> October 7 1999
5 Bodo Bauer <bb@ricochet.net>
6
72.4.x update Jorge Nerin <comandante@zaralinux.com> November 14 2000
Stefani Seibold349888e2009-06-17 16:26:01 -07008move /proc/sys Shen Feng <shen@cn.fujitsu.com> April 1 2009
Linus Torvalds1da177e2005-04-16 15:20:36 -07009------------------------------------------------------------------------------
10Version 1.3 Kernel version 2.2.12
11 Kernel version 2.4.0-test11-pre4
12------------------------------------------------------------------------------
Stefani Seibold349888e2009-06-17 16:26:01 -070013fixes/update part 1.1 Stefani Seibold <stefani@seibold.net> June 9 2009
Linus Torvalds1da177e2005-04-16 15:20:36 -070014
15Table of Contents
16-----------------
17
18 0 Preface
19 0.1 Introduction/Credits
20 0.2 Legal Stuff
21
22 1 Collecting System Information
23 1.1 Process-Specific Subdirectories
24 1.2 Kernel data
25 1.3 IDE devices in /proc/ide
26 1.4 Networking info in /proc/net
27 1.5 SCSI info
28 1.6 Parallel port info in /proc/parport
29 1.7 TTY info in /proc/tty
30 1.8 Miscellaneous kernel statistics in /proc/stat
Trace Pillarsae96b342015-01-23 11:45:05 -050031 1.9 Ext4 file system parameters
Linus Torvalds1da177e2005-04-16 15:20:36 -070032
33 2 Modifying System Parameters
Shen Feng760df932009-04-02 16:57:20 -070034
35 3 Per-Process Parameters
David Rientjesfa0cbbf2012-11-12 17:53:04 -080036 3.1 /proc/<pid>/oom_adj & /proc/<pid>/oom_score_adj - Adjust the oom-killer
David Rientjesa63d83f2010-08-09 17:19:46 -070037 score
Shen Feng760df932009-04-02 16:57:20 -070038 3.2 /proc/<pid>/oom_score - Display current oom-killer score
39 3.3 /proc/<pid>/io - Display the IO accounting fields
40 3.4 /proc/<pid>/coredump_filter - Core dump filtering settings
41 3.5 /proc/<pid>/mountinfo - Information about mounts
john stultz4614a696b2009-12-14 18:00:05 -080042 3.6 /proc/<pid>/comm & /proc/<pid>/task/<tid>/comm
Cyrill Gorcunov818411612012-05-31 16:26:43 -070043 3.7 /proc/<pid>/task/<tid>/children - Information about task children
Cyrill Gorcunovf1d8c162012-12-17 16:05:14 -080044 3.8 /proc/<pid>/fdinfo/<fd> - Information about opened file
Cyrill Gorcunov740a5dd2015-02-11 15:28:31 -080045 3.9 /proc/<pid>/map_files - Information about memory mapped files
John Stultz5de23d42016-03-17 14:20:54 -070046 3.10 /proc/<pid>/timerslack_ns - Task timerslack value
Josh Poimboeuf7c23b332017-02-13 19:42:41 -060047 3.11 /proc/<pid>/patch_state - Livepatch patch operation state
Shen Feng760df932009-04-02 16:57:20 -070048
Vasiliy Kulikov04996802012-01-10 15:11:31 -080049 4 Configuring procfs
50 4.1 Mount options
Linus Torvalds1da177e2005-04-16 15:20:36 -070051
52------------------------------------------------------------------------------
53Preface
54------------------------------------------------------------------------------
55
560.1 Introduction/Credits
57------------------------
58
59This documentation is part of a soon (or so we hope) to be released book on
60the SuSE Linux distribution. As there is no complete documentation for the
61/proc file system and we've used many freely available sources to write these
62chapters, it seems only fair to give the work back to the Linux community.
63This work is based on the 2.2.* kernel version and the upcoming 2.4.*. I'm
64afraid it's still far from complete, but we hope it will be useful. As far as
65we know, it is the first 'all-in-one' document about the /proc file system. It
66is focused on the Intel x86 hardware, so if you are looking for PPC, ARM,
67SPARC, AXP, etc., features, you probably won't find what you are looking for.
68It also only covers IPv4 networking, not IPv6 nor other protocols - sorry. But
69additions and patches are welcome and will be added to this document if you
70mail them to Bodo.
71
72We'd like to thank Alan Cox, Rik van Riel, and Alexey Kuznetsov and a lot of
73other people for help compiling this documentation. We'd also like to extend a
74special thank you to Andi Kleen for documentation, which we relied on heavily
75to create this document, as well as the additional information he provided.
76Thanks to everybody else who contributed source or docs to the Linux kernel
77and helped create a great piece of software... :)
78
79If you have any comments, corrections or additions, please don't hesitate to
80contact Bodo Bauer at bb@ricochet.net. We'll be happy to add them to this
81document.
82
83The latest version of this document is available online at
Justin P. Mattock0ea6e612010-07-23 20:51:24 -070084http://tldp.org/LDP/Linux-Filesystem-Hierarchy/html/proc.html
Linus Torvalds1da177e2005-04-16 15:20:36 -070085
Justin P. Mattock0ea6e612010-07-23 20:51:24 -070086If the above direction does not works for you, you could try the kernel
Linus Torvalds1da177e2005-04-16 15:20:36 -070087mailing list at linux-kernel@vger.kernel.org and/or try to reach me at
88comandante@zaralinux.com.
89
900.2 Legal Stuff
91---------------
92
93We don't guarantee the correctness of this document, and if you come to us
94complaining about how you screwed up your system because of incorrect
95documentation, we won't feel responsible...
96
97------------------------------------------------------------------------------
98CHAPTER 1: COLLECTING SYSTEM INFORMATION
99------------------------------------------------------------------------------
100
101------------------------------------------------------------------------------
102In This Chapter
103------------------------------------------------------------------------------
104* Investigating the properties of the pseudo file system /proc and its
105 ability to provide information on the running Linux system
106* Examining /proc's structure
107* Uncovering various information about the kernel and the processes running
108 on the system
109------------------------------------------------------------------------------
110
111
112The proc file system acts as an interface to internal data structures in the
113kernel. It can be used to obtain information about the system and to change
114certain kernel parameters at runtime (sysctl).
115
116First, we'll take a look at the read-only parts of /proc. In Chapter 2, we
117show you how you can use /proc/sys to change settings.
118
1191.1 Process-Specific Subdirectories
120-----------------------------------
121
122The directory /proc contains (among other things) one subdirectory for each
123process running on the system, which is named after the process ID (PID).
124
125The link self points to the process reading the file system. Each process
126subdirectory has the entries listed in Table 1-1.
127
Daniel Colascionec969eb82018-11-05 13:22:05 +0000128Note that an open a file descriptor to /proc/<pid> or to any of its
129contained files or subdirectories does not prevent <pid> being reused
130for some other process in the event that <pid> exits. Operations on
131open /proc/<pid> file descriptors corresponding to dead processes
132never act on any new process that the kernel may, through chance, have
133also assigned the process ID <pid>. Instead, operations on these FDs
134usually fail with ESRCH.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700135
Stefani Seibold349888e2009-06-17 16:26:01 -0700136Table 1-1: Process specific entries in /proc
Linus Torvalds1da177e2005-04-16 15:20:36 -0700137..............................................................................
David Rientjesb813e932007-05-06 14:49:24 -0700138 File Content
139 clear_refs Clears page referenced bits shown in smaps output
140 cmdline Command line arguments
141 cpu Current and last cpu in which it was executed (2.4)(smp)
142 cwd Link to the current working directory
143 environ Values of environment variables
144 exe Link to the executable of this process
145 fd Directory, which contains all file descriptors
146 maps Memory maps to executables and library files (2.4)
147 mem Memory held by this process
148 root Link to the root directory of this process
149 stat Process status
150 statm Process memory status information
151 status Process status in human readable form
Ingo Molnarb2f73922015-09-30 15:59:17 +0200152 wchan Present with CONFIG_KALLSYMS=y: it shows the kernel function
153 symbol the task is blocked in - or "0" if not blocked.
Nikanth Karthikesan03f890f2010-10-27 15:34:11 -0700154 pagemap Page table
Ken Chen2ec220e2008-11-10 11:26:08 +0300155 stack Report full stack trace, enable via CONFIG_STACKTRACE
Robert Foss3d8819b2016-09-08 18:44:23 -0400156 smaps an extension based on maps, showing the memory consumption of
Cyrill Gorcunov834f82e2012-12-17 16:03:13 -0800157 each mapping and flags associated with it
Rafael Aquini0c369712015-02-12 15:01:05 -0800158 numa_maps an extension based on maps, showing the memory locality and
159 binding policy as well as mem usage (in pages) of each mapping.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700160..............................................................................
161
162For example, to get the status information of a process, all you have to do is
163read the file /proc/PID/status:
164
Stefani Seibold349888e2009-06-17 16:26:01 -0700165 >cat /proc/self/status
166 Name: cat
167 State: R (running)
168 Tgid: 5452
169 Pid: 5452
170 PPid: 743
Linus Torvalds1da177e2005-04-16 15:20:36 -0700171 TracerPid: 0 (2.4)
Stefani Seibold349888e2009-06-17 16:26:01 -0700172 Uid: 501 501 501 501
173 Gid: 100 100 100 100
174 FDSize: 256
175 Groups: 100 14 16
176 VmPeak: 5004 kB
177 VmSize: 5004 kB
178 VmLck: 0 kB
179 VmHWM: 476 kB
180 VmRSS: 476 kB
Jerome Marchand8cee8522016-01-14 15:19:29 -0800181 RssAnon: 352 kB
182 RssFile: 120 kB
183 RssShmem: 4 kB
Stefani Seibold349888e2009-06-17 16:26:01 -0700184 VmData: 156 kB
185 VmStk: 88 kB
186 VmExe: 68 kB
187 VmLib: 1412 kB
188 VmPTE: 20 kb
KAMEZAWA Hiroyukib084d432010-03-05 13:41:42 -0800189 VmSwap: 0 kB
Naoya Horiguchi5d317b22015-11-05 18:47:14 -0800190 HugetlbPages: 0 kB
Roman Gushchinc6434012017-11-17 15:26:45 -0800191 CoreDumping: 0
Michal Hockoa1400af2018-12-28 00:38:25 -0800192 THP_enabled: 1
Stefani Seibold349888e2009-06-17 16:26:01 -0700193 Threads: 1
194 SigQ: 0/28578
195 SigPnd: 0000000000000000
196 ShdPnd: 0000000000000000
197 SigBlk: 0000000000000000
198 SigIgn: 0000000000000000
199 SigCgt: 0000000000000000
200 CapInh: 00000000fffffeff
201 CapPrm: 0000000000000000
202 CapEff: 0000000000000000
203 CapBnd: ffffffffffffffff
Waiman Longf8d0dc22018-10-23 17:25:51 -0400204 CapAmb: 0000000000000000
Kees Cookaf884cd2016-12-12 16:45:05 -0800205 NoNewPrivs: 0
Kees Cook2f4b3bf2012-12-17 16:03:14 -0800206 Seccomp: 0
Waiman Longf8d0dc22018-10-23 17:25:51 -0400207 Speculation_Store_Bypass: thread vulnerable
Stefani Seibold349888e2009-06-17 16:26:01 -0700208 voluntary_ctxt_switches: 0
209 nonvoluntary_ctxt_switches: 1
Linus Torvalds1da177e2005-04-16 15:20:36 -0700210
211This shows you nearly the same information you would get if you viewed it with
212the ps command. In fact, ps uses the proc file system to obtain its
Stefani Seibold349888e2009-06-17 16:26:01 -0700213information. But you get a more detailed view of the process by reading the
214file /proc/PID/status. It fields are described in table 1-2.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700215
Stefani Seibold349888e2009-06-17 16:26:01 -0700216The statm file contains more detailed information about the process
217memory usage. Its seven fields are explained in Table 1-3. The stat file
218contains details information about the process itself. Its fields are
219explained in Table 1-4.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700220
KAMEZAWA Hiroyuki34e55232010-03-05 13:41:40 -0800221(for SMP CONFIG users)
Nathan Scott15eb42d2015-04-16 12:49:35 -0700222For making accounting scalable, RSS related information are handled in an
223asynchronous manner and the value may not be very precise. To see a precise
KAMEZAWA Hiroyuki34e55232010-03-05 13:41:40 -0800224snapshot of a moment, you can see /proc/<pid>/smaps file and scan page table.
225It's slow but very precise.
226
Waiman Longf8d0dc22018-10-23 17:25:51 -0400227Table 1-2: Contents of the status files (as of 4.19)
Stefani Seibold349888e2009-06-17 16:26:01 -0700228..............................................................................
229 Field Content
230 Name filename of the executable
Fabian Frederickbbd88e12017-01-24 15:18:13 -0800231 Umask file mode creation mask
Stefani Seibold349888e2009-06-17 16:26:01 -0700232 State state (R is running, S is sleeping, D is sleeping
233 in an uninterruptible wait, Z is zombie,
234 T is traced or stopped)
235 Tgid thread group ID
Nathan Scott15eb42d2015-04-16 12:49:35 -0700236 Ngid NUMA group ID (0 if none)
Stefani Seibold349888e2009-06-17 16:26:01 -0700237 Pid process id
238 PPid process id of the parent process
239 TracerPid PID of process tracing this process (0 if not)
240 Uid Real, effective, saved set, and file system UIDs
241 Gid Real, effective, saved set, and file system GIDs
242 FDSize number of file descriptor slots currently allocated
243 Groups supplementary group list
Nathan Scott15eb42d2015-04-16 12:49:35 -0700244 NStgid descendant namespace thread group ID hierarchy
245 NSpid descendant namespace process ID hierarchy
246 NSpgid descendant namespace process group ID hierarchy
247 NSsid descendant namespace session ID hierarchy
Stefani Seibold349888e2009-06-17 16:26:01 -0700248 VmPeak peak virtual memory size
249 VmSize total program size
250 VmLck locked memory size
Fabian Frederickbbd88e12017-01-24 15:18:13 -0800251 VmPin pinned memory size
Stefani Seibold349888e2009-06-17 16:26:01 -0700252 VmHWM peak resident set size ("high water mark")
Jerome Marchand8cee8522016-01-14 15:19:29 -0800253 VmRSS size of memory portions. It contains the three
254 following parts (VmRSS = RssAnon + RssFile + RssShmem)
255 RssAnon size of resident anonymous memory
256 RssFile size of resident file mappings
257 RssShmem size of resident shmem memory (includes SysV shm,
258 mapping of tmpfs and shared anonymous mappings)
Konstantin Khlebnikov30bdbb72016-02-02 16:57:46 -0800259 VmData size of private data segments
260 VmStk size of stack segments
Stefani Seibold349888e2009-06-17 16:26:01 -0700261 VmExe size of text segment
262 VmLib size of shared library code
263 VmPTE size of page table entries
Vlastimil Babkabf9683d2016-01-14 15:19:14 -0800264 VmSwap amount of swap used by anonymous private data
265 (shmem swap usage is not included)
Naoya Horiguchi5d317b22015-11-05 18:47:14 -0800266 HugetlbPages size of hugetlb memory portions
Roman Gushchinc6434012017-11-17 15:26:45 -0800267 CoreDumping process's memory is currently being dumped
268 (killing the process may lead to a corrupted core)
Michal Hockoa1400af2018-12-28 00:38:25 -0800269 THP_enabled process is allowed to use THP (returns 0 when
270 PR_SET_THP_DISABLE is set on the process
Stefani Seibold349888e2009-06-17 16:26:01 -0700271 Threads number of threads
272 SigQ number of signals queued/max. number for queue
273 SigPnd bitmap of pending signals for the thread
274 ShdPnd bitmap of shared pending signals for the process
275 SigBlk bitmap of blocked signals
276 SigIgn bitmap of ignored signals
Carlos Garciac98be0c2014-04-04 22:31:00 -0400277 SigCgt bitmap of caught signals
Stefani Seibold349888e2009-06-17 16:26:01 -0700278 CapInh bitmap of inheritable capabilities
279 CapPrm bitmap of permitted capabilities
280 CapEff bitmap of effective capabilities
281 CapBnd bitmap of capabilities bounding set
Waiman Longf8d0dc22018-10-23 17:25:51 -0400282 CapAmb bitmap of ambient capabilities
Kees Cookaf884cd2016-12-12 16:45:05 -0800283 NoNewPrivs no_new_privs, like prctl(PR_GET_NO_NEW_PRIV, ...)
Kees Cook2f4b3bf2012-12-17 16:03:14 -0800284 Seccomp seccomp mode, like prctl(PR_GET_SECCOMP, ...)
Waiman Longf8d0dc22018-10-23 17:25:51 -0400285 Speculation_Store_Bypass speculative store bypass mitigation status
Stefani Seibold349888e2009-06-17 16:26:01 -0700286 Cpus_allowed mask of CPUs on which this process may run
287 Cpus_allowed_list Same as previous, but in "list format"
288 Mems_allowed mask of memory nodes allowed to this process
289 Mems_allowed_list Same as previous, but in "list format"
290 voluntary_ctxt_switches number of voluntary context switches
291 nonvoluntary_ctxt_switches number of non voluntary context switches
292..............................................................................
293
294Table 1-3: Contents of the statm files (as of 2.6.8-rc3)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700295..............................................................................
296 Field Content
297 size total program size (pages) (same as VmSize in status)
298 resident size of memory portions (pages) (same as VmRSS in status)
Jerome Marchand8cee8522016-01-14 15:19:29 -0800299 shared number of pages that are shared (i.e. backed by a file, same
300 as RssFile+RssShmem in status)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700301 trs number of pages that are 'code' (not including libs; broken,
302 includes data segment)
303 lrs number of pages of library (always 0 on 2.6)
304 drs number of pages of data/stack (including libs; broken,
305 includes library text)
306 dt number of dirty pages (always 0 on 2.6)
307..............................................................................
308
Kees Cook18d96772007-07-15 23:40:38 -0700309
Stefani Seibold349888e2009-06-17 16:26:01 -0700310Table 1-4: Contents of the stat files (as of 2.6.30-rc7)
Kees Cook18d96772007-07-15 23:40:38 -0700311..............................................................................
312 Field Content
313 pid process id
314 tcomm filename of the executable
315 state state (R is running, S is sleeping, D is sleeping in an
316 uninterruptible wait, Z is zombie, T is traced or stopped)
317 ppid process id of the parent process
318 pgrp pgrp of the process
319 sid session id
320 tty_nr tty the process uses
321 tty_pgrp pgrp of the tty
322 flags task flags
323 min_flt number of minor faults
324 cmin_flt number of minor faults with child's
325 maj_flt number of major faults
326 cmaj_flt number of major faults with child's
327 utime user mode jiffies
328 stime kernel mode jiffies
329 cutime user mode jiffies with child's
330 cstime kernel mode jiffies with child's
331 priority priority level
332 nice nice level
333 num_threads number of threads
Leonardo Chiquitto2e01e002008-02-03 16:17:16 +0200334 it_real_value (obsolete, always 0)
Kees Cook18d96772007-07-15 23:40:38 -0700335 start_time time the process started after system boot
336 vsize virtual memory size
337 rss resident set memory size
338 rsslim current limit in bytes on the rss
339 start_code address above which program text can run
340 end_code address below which program text can run
Siddhesh Poyarekarb7643752012-03-21 16:34:04 -0700341 start_stack address of the start of the main process stack
Kees Cook18d96772007-07-15 23:40:38 -0700342 esp current value of ESP
343 eip current value of EIP
Stefani Seibold349888e2009-06-17 16:26:01 -0700344 pending bitmap of pending signals
345 blocked bitmap of blocked signals
346 sigign bitmap of ignored signals
Carlos Garciac98be0c2014-04-04 22:31:00 -0400347 sigcatch bitmap of caught signals
Ingo Molnarb2f73922015-09-30 15:59:17 +0200348 0 (place holder, used to be the wchan address, use /proc/PID/wchan instead)
Kees Cook18d96772007-07-15 23:40:38 -0700349 0 (place holder)
350 0 (place holder)
351 exit_signal signal to send to parent thread on exit
352 task_cpu which CPU the task is scheduled on
353 rt_priority realtime priority
354 policy scheduling policy (man sched_setscheduler)
355 blkio_ticks time spent waiting for block IO
Stefani Seibold349888e2009-06-17 16:26:01 -0700356 gtime guest time of the task in jiffies
357 cgtime guest time of the task children in jiffies
Cyrill Gorcunovb3f7f572012-01-12 17:20:53 -0800358 start_data address above which program data+bss is placed
359 end_data address below which program data+bss is placed
360 start_brk address above which program heap can be expanded with brk()
Cyrill Gorcunov5b172082012-05-31 16:26:44 -0700361 arg_start address above which program command line is placed
362 arg_end address below which program command line is placed
363 env_start address above which program environment is placed
364 env_end address below which program environment is placed
365 exit_code the thread's exit_code in the form reported by the waitpid system call
Kees Cook18d96772007-07-15 23:40:38 -0700366..............................................................................
367
Rob Landley32e688b2010-03-15 15:21:31 +0100368The /proc/PID/maps file containing the currently mapped memory regions and
Stefani Seibold349888e2009-06-17 16:26:01 -0700369their access permissions.
370
371The format is:
372
373address perms offset dev inode pathname
374
37508048000-08049000 r-xp 00000000 03:00 8312 /opt/test
37608049000-0804a000 rw-p 00001000 03:00 8312 /opt/test
3770804a000-0806b000 rw-p 00000000 00:00 0 [heap]
378a7cb1000-a7cb2000 ---p 00000000 00:00 0
Robin Holt34441422010-05-11 14:06:46 -0700379a7cb2000-a7eb2000 rw-p 00000000 00:00 0
Stefani Seibold349888e2009-06-17 16:26:01 -0700380a7eb2000-a7eb3000 ---p 00000000 00:00 0
Johannes Weiner65376df2016-02-02 16:57:29 -0800381a7eb3000-a7ed5000 rw-p 00000000 00:00 0
Stefani Seibold349888e2009-06-17 16:26:01 -0700382a7ed5000-a8008000 r-xp 00000000 03:00 4222 /lib/libc.so.6
383a8008000-a800a000 r--p 00133000 03:00 4222 /lib/libc.so.6
384a800a000-a800b000 rw-p 00135000 03:00 4222 /lib/libc.so.6
385a800b000-a800e000 rw-p 00000000 00:00 0
386a800e000-a8022000 r-xp 00000000 03:00 14462 /lib/libpthread.so.0
387a8022000-a8023000 r--p 00013000 03:00 14462 /lib/libpthread.so.0
388a8023000-a8024000 rw-p 00014000 03:00 14462 /lib/libpthread.so.0
389a8024000-a8027000 rw-p 00000000 00:00 0
390a8027000-a8043000 r-xp 00000000 03:00 8317 /lib/ld-linux.so.2
391a8043000-a8044000 r--p 0001b000 03:00 8317 /lib/ld-linux.so.2
392a8044000-a8045000 rw-p 0001c000 03:00 8317 /lib/ld-linux.so.2
393aff35000-aff4a000 rw-p 00000000 00:00 0 [stack]
394ffffe000-fffff000 r-xp 00000000 00:00 0 [vdso]
395
396where "address" is the address space in the process that it occupies, "perms"
397is a set of permissions:
398
399 r = read
400 w = write
401 x = execute
402 s = shared
403 p = private (copy on write)
404
405"offset" is the offset into the mapping, "dev" is the device (major:minor), and
406"inode" is the inode on that device. 0 indicates that no inode is associated
407with the memory region, as the case would be with BSS (uninitialized data).
408The "pathname" shows the name associated file for this mapping. If the mapping
409is not associated with a file:
410
411 [heap] = the heap of the program
412 [stack] = the stack of the main process
413 [vdso] = the "virtual dynamic shared object",
414 the kernel system call handler
415
416 or if empty, the mapping is anonymous.
417
Stefani Seibold349888e2009-06-17 16:26:01 -0700418The /proc/PID/smaps is an extension based on maps, showing the memory
419consumption for each of the process's mappings. For each of mappings there
420is a series of lines such as the following:
421
42208048000-080bc000 r-xp 00000000 03:02 13130 /bin/bash
423Size: 1084 kB
424Rss: 892 kB
425Pss: 374 kB
426Shared_Clean: 892 kB
427Shared_Dirty: 0 kB
428Private_Clean: 0 kB
429Private_Dirty: 0 kB
430Referenced: 892 kB
Nikanth Karthikesanb40d4f82010-10-27 15:34:10 -0700431Anonymous: 0 kB
Shaohua Licf8496e2017-05-03 14:52:42 -0700432LazyFree: 0 kB
Naoya Horiguchi25ee01a2015-11-05 18:47:11 -0800433AnonHugePages: 0 kB
Kirill A. Shutemov1b5946a2016-07-26 15:26:40 -0700434ShmemPmdMapped: 0 kB
Naoya Horiguchi25ee01a2015-11-05 18:47:11 -0800435Shared_Hugetlb: 0 kB
436Private_Hugetlb: 0 kB
Stefani Seibold349888e2009-06-17 16:26:01 -0700437Swap: 0 kB
Minchan Kim8334b962015-09-08 15:00:24 -0700438SwapPss: 0 kB
Stefani Seibold349888e2009-06-17 16:26:01 -0700439KernelPageSize: 4 kB
440MMUPageSize: 4 kB
Hugh Dickinsa5be3562015-11-05 18:50:37 -0800441Locked: 0 kB
Michal Hocko7635d9c2018-12-28 00:38:21 -0800442THPeligible: 0
Hugh Dickinsa5be3562015-11-05 18:50:37 -0800443VmFlags: rd ex mr mw me dw
Stefani Seibold349888e2009-06-17 16:26:01 -0700444
Cyrill Gorcunov834f82e2012-12-17 16:03:13 -0800445the first of these lines shows the same information as is displayed for the
Matt Mackall0f4d2082010-10-26 14:21:22 -0700446mapping in /proc/PID/maps. The remaining lines show the size of the mapping
447(size), the amount of the mapping that is currently resident in RAM (RSS), the
448process' proportional share of this mapping (PSS), the number of clean and
Minchan Kim8334b962015-09-08 15:00:24 -0700449dirty private pages in the mapping.
450
451The "proportional set size" (PSS) of a process is the count of pages it has
452in memory, where each page is divided by the number of processes sharing it.
453So if a process has 1000 pages all to itself, and 1000 shared with one other
454process, its PSS will be 1500.
455Note that even a page which is part of a MAP_SHARED mapping, but has only
456a single pte mapped, i.e. is currently used by only one process, is accounted
457as private and not as shared.
458"Referenced" indicates the amount of memory currently marked as referenced or
459accessed.
Nikanth Karthikesanb40d4f82010-10-27 15:34:10 -0700460"Anonymous" shows the amount of memory that does not belong to any file. Even
461a mapping associated with a file may contain anonymous pages: when MAP_PRIVATE
462and a page is modified, the file page is replaced by a private anonymous copy.
Shaohua Licf8496e2017-05-03 14:52:42 -0700463"LazyFree" shows the amount of memory which is marked by madvise(MADV_FREE).
464The memory isn't freed immediately with madvise(). It's freed in memory
465pressure if the memory is clean. Please note that the printed value might
466be lower than the real value due to optimizations used in the current
467implementation. If this is not desirable please file a bug report.
Naoya Horiguchi25ee01a2015-11-05 18:47:11 -0800468"AnonHugePages" shows the ammount of memory backed by transparent hugepage.
Kirill A. Shutemov1b5946a2016-07-26 15:26:40 -0700469"ShmemPmdMapped" shows the ammount of shared (shmem/tmpfs) memory backed by
470huge pages.
Naoya Horiguchi25ee01a2015-11-05 18:47:11 -0800471"Shared_Hugetlb" and "Private_Hugetlb" show the ammounts of memory backed by
472hugetlbfs page which is *not* counted in "RSS" or "PSS" field for historical
473reasons. And these are not included in {Shared,Private}_{Clean,Dirty} field.
Hugh Dickinsa5be3562015-11-05 18:50:37 -0800474"Swap" shows how much would-be-anonymous memory is also used, but out on swap.
Vlastimil Babkac261e7d92016-01-14 15:19:17 -0800475For shmem mappings, "Swap" includes also the size of the mapped (and not
476replaced by copy-on-write) part of the underlying shmem object out on swap.
477"SwapPss" shows proportional swap share of this mapping. Unlike "Swap", this
478does not take into account swapped out page of underlying shmem objects.
Hugh Dickinsa5be3562015-11-05 18:50:37 -0800479"Locked" indicates whether the mapping is locked in memory or not.
Michal Hocko7635d9c2018-12-28 00:38:21 -0800480"THPeligible" indicates whether the mapping is eligible for THP pages - 1 if
481true, 0 otherwise.
Naoya Horiguchi25ee01a2015-11-05 18:47:11 -0800482
Cyrill Gorcunov834f82e2012-12-17 16:03:13 -0800483"VmFlags" field deserves a separate description. This member represents the kernel
484flags associated with the particular virtual memory area in two letter encoded
485manner. The codes are the following:
486 rd - readable
487 wr - writeable
488 ex - executable
489 sh - shared
490 mr - may read
491 mw - may write
492 me - may execute
493 ms - may share
494 gd - stack segment growns down
495 pf - pure PFN range
496 dw - disabled write to the mapped file
497 lo - pages are locked in memory
498 io - memory mapped I/O area
499 sr - sequential read advise provided
500 rr - random read advise provided
501 dc - do not copy area on fork
502 de - do not expand area on remapping
503 ac - area is accountable
504 nr - swap space is not reserved for the area
505 ht - area uses huge tlb pages
Cyrill Gorcunov834f82e2012-12-17 16:03:13 -0800506 ar - architecture specific flag
507 dd - do not include area into core dump
Naoya Horiguchiec8e41a2013-11-12 15:07:49 -0800508 sd - soft-dirty flag
Cyrill Gorcunov834f82e2012-12-17 16:03:13 -0800509 mm - mixed map area
510 hg - huge page advise flag
511 nh - no-huge page advise flag
512 mg - mergable advise flag
513
514Note that there is no guarantee that every flag and associated mnemonic will
515be present in all further kernel releases. Things get changed, the flags may
Michal Hocko7550c602018-12-28 00:38:17 -0800516be vanished or the reverse -- new added. Interpretation of their meaning
517might change in future as well. So each consumer of these flags has to
518follow each specific kernel version for the exact semantic.
Cyrill Gorcunov834f82e2012-12-17 16:03:13 -0800519
Stefani Seibold349888e2009-06-17 16:26:01 -0700520This file is only present if the CONFIG_MMU kernel configuration option is
521enabled.
Kees Cook18d96772007-07-15 23:40:38 -0700522
Robert Ho53aeee72016-10-07 17:02:39 -0700523Note: reading /proc/PID/maps or /proc/PID/smaps is inherently racy (consistent
524output can be achieved only in the single read call).
525This typically manifests when doing partial reads of these files while the
526memory map is being modified. Despite the races, we do provide the following
527guarantees:
528
5291) The mapped addresses never go backwards, which implies no two
530 regions will ever overlap.
5312) If there is something at a given vaddr during the entirety of the
532 life of the smaps/maps walk, there will be some output for it.
533
534
Moussa A. Ba398499d2009-09-21 17:02:29 -0700535The /proc/PID/clear_refs is used to reset the PG_Referenced and ACCESSED/YOUNG
Pavel Emelyanov0f8975e2013-07-03 15:01:20 -0700536bits on both physical and virtual pages associated with a process, and the
Mike Rapoport1ad13352018-04-18 11:07:49 +0300537soft-dirty bit on pte (see Documentation/admin-guide/mm/soft-dirty.rst
538for details).
Moussa A. Ba398499d2009-09-21 17:02:29 -0700539To clear the bits for all the pages associated with the process
540 > echo 1 > /proc/PID/clear_refs
541
542To clear the bits for the anonymous pages associated with the process
543 > echo 2 > /proc/PID/clear_refs
544
545To clear the bits for the file mapped pages associated with the process
546 > echo 3 > /proc/PID/clear_refs
Pavel Emelyanov0f8975e2013-07-03 15:01:20 -0700547
548To clear the soft-dirty bit
549 > echo 4 > /proc/PID/clear_refs
550
Petr Cermak695f0552015-02-12 15:01:00 -0800551To reset the peak resident set size ("high water mark") to the process's
552current value:
553 > echo 5 > /proc/PID/clear_refs
554
Moussa A. Ba398499d2009-09-21 17:02:29 -0700555Any other value written to /proc/PID/clear_refs will have no effect.
556
Nikanth Karthikesan03f890f2010-10-27 15:34:11 -0700557The /proc/pid/pagemap gives the PFN, which can be used to find the pageflags
558using /proc/kpageflags and number of times a page is mapped using
Mike Rapoport1ad13352018-04-18 11:07:49 +0300559/proc/kpagecount. For detailed explanation, see
560Documentation/admin-guide/mm/pagemap.rst.
Moussa A. Ba398499d2009-09-21 17:02:29 -0700561
Rafael Aquini0c369712015-02-12 15:01:05 -0800562The /proc/pid/numa_maps is an extension based on maps, showing the memory
563locality and binding policy, as well as the memory usage (in pages) of
564each mapping. The output follows a general format where mapping details get
565summarized separated by blank spaces, one mapping per each file line:
566
567address policy mapping details
568
Rafael Aquini198d1592015-02-12 15:01:08 -080056900400000 default file=/usr/local/bin/app mapped=1 active=0 N3=1 kernelpagesize_kB=4
57000600000 default file=/usr/local/bin/app anon=1 dirty=1 N3=1 kernelpagesize_kB=4
5713206000000 default file=/lib64/ld-2.12.so mapped=26 mapmax=6 N0=24 N3=2 kernelpagesize_kB=4
572320621f000 default file=/lib64/ld-2.12.so anon=1 dirty=1 N3=1 kernelpagesize_kB=4
5733206220000 default file=/lib64/ld-2.12.so anon=1 dirty=1 N3=1 kernelpagesize_kB=4
5743206221000 default anon=1 dirty=1 N3=1 kernelpagesize_kB=4
5753206800000 default file=/lib64/libc-2.12.so mapped=59 mapmax=21 active=55 N0=41 N3=18 kernelpagesize_kB=4
Rafael Aquini0c369712015-02-12 15:01:05 -0800576320698b000 default file=/lib64/libc-2.12.so
Rafael Aquini198d1592015-02-12 15:01:08 -08005773206b8a000 default file=/lib64/libc-2.12.so anon=2 dirty=2 N3=2 kernelpagesize_kB=4
5783206b8e000 default file=/lib64/libc-2.12.so anon=1 dirty=1 N3=1 kernelpagesize_kB=4
5793206b8f000 default anon=3 dirty=3 active=1 N3=3 kernelpagesize_kB=4
5807f4dc10a2000 default anon=3 dirty=3 N3=3 kernelpagesize_kB=4
5817f4dc10b4000 default anon=2 dirty=2 active=1 N3=2 kernelpagesize_kB=4
5827f4dc1200000 default file=/anon_hugepage\040(deleted) huge anon=1 dirty=1 N3=1 kernelpagesize_kB=2048
5837fff335f0000 default stack anon=3 dirty=3 N3=3 kernelpagesize_kB=4
5847fff3369d000 default mapped=1 mapmax=35 active=0 N3=1 kernelpagesize_kB=4
Rafael Aquini0c369712015-02-12 15:01:05 -0800585
586Where:
587"address" is the starting address for the mapping;
Mike Rapoport3ecf53e2018-05-08 10:02:10 +0300588"policy" reports the NUMA memory policy set for the mapping (see Documentation/admin-guide/mm/numa_memory_policy.rst);
Rafael Aquini0c369712015-02-12 15:01:05 -0800589"mapping details" summarizes mapping data such as mapping type, page usage counters,
590node locality page counters (N0 == node0, N1 == node1, ...) and the kernel page
591size, in KB, that is backing the mapping up.
592
Linus Torvalds1da177e2005-04-16 15:20:36 -07005931.2 Kernel data
594---------------
595
596Similar to the process entries, the kernel data files give information about
597the running kernel. The files used to obtain this information are contained in
Stefani Seibold349888e2009-06-17 16:26:01 -0700598/proc and are listed in Table 1-5. Not all of these will be present in your
Linus Torvalds1da177e2005-04-16 15:20:36 -0700599system. It depends on the kernel configuration and the loaded modules, which
600files are there, and which are missing.
601
Stefani Seibold349888e2009-06-17 16:26:01 -0700602Table 1-5: Kernel info in /proc
Linus Torvalds1da177e2005-04-16 15:20:36 -0700603..............................................................................
604 File Content
605 apm Advanced power management info
606 buddyinfo Kernel memory allocator information (see text) (2.5)
607 bus Directory containing bus specific information
608 cmdline Kernel command line
609 cpuinfo Info about the CPU
610 devices Available devices (block and character)
611 dma Used DMS channels
612 filesystems Supported filesystems
613 driver Various drivers grouped here, currently rtc (2.4)
614 execdomains Execdomains, related to security (2.4)
615 fb Frame Buffer devices (2.4)
616 fs File system parameters, currently nfs/exports (2.4)
617 ide Directory containing info about the IDE subsystem
618 interrupts Interrupt usage
619 iomem Memory map (2.4)
620 ioports I/O port usage
621 irq Masks for irq to cpu affinity (2.4)(smp?)
622 isapnp ISA PnP (Plug&Play) Info (2.4)
623 kcore Kernel core image (can be ELF or A.OUT(deprecated in 2.4))
624 kmsg Kernel messages
625 ksyms Kernel symbol table
626 loadavg Load average of last 1, 5 & 15 minutes
627 locks Kernel locks
628 meminfo Memory info
629 misc Miscellaneous
630 modules List of loaded modules
631 mounts Mounted filesystems
632 net Networking info (see text)
Mel Gormana1b57ac2010-03-05 13:42:15 -0800633 pagetypeinfo Additional page allocator information (see text) (2.5)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700634 partitions Table of partitions known to the system
Randy Dunlap8b607562007-05-09 07:19:14 +0200635 pci Deprecated info of PCI bus (new way -> /proc/bus/pci/,
Linus Torvalds1da177e2005-04-16 15:20:36 -0700636 decoupled by lspci (2.4)
637 rtc Real time clock
638 scsi SCSI info (see text)
639 slabinfo Slab pool info
Keika Kobayashid3d64df2009-06-17 16:25:55 -0700640 softirqs softirq usage
Linus Torvalds1da177e2005-04-16 15:20:36 -0700641 stat Overall statistics
642 swaps Swap space utilization
643 sys See chapter 2
644 sysvipc Info of SysVIPC Resources (msg, sem, shm) (2.4)
645 tty Info of tty drivers
Rob Landley49457892013-12-31 22:34:04 -0600646 uptime Wall clock since boot, combined idle time of all cpus
Linus Torvalds1da177e2005-04-16 15:20:36 -0700647 version Kernel version
648 video bttv info of video resources (2.4)
Eric Dumazeta47a1262008-07-23 21:27:38 -0700649 vmallocinfo Show vmalloced areas
Linus Torvalds1da177e2005-04-16 15:20:36 -0700650..............................................................................
651
652You can, for example, check which interrupts are currently in use and what
653they are used for by looking in the file /proc/interrupts:
654
655 > cat /proc/interrupts
656 CPU0
657 0: 8728810 XT-PIC timer
658 1: 895 XT-PIC keyboard
659 2: 0 XT-PIC cascade
660 3: 531695 XT-PIC aha152x
661 4: 2014133 XT-PIC serial
662 5: 44401 XT-PIC pcnet_cs
663 8: 2 XT-PIC rtc
664 11: 8 XT-PIC i82365
665 12: 182918 XT-PIC PS/2 Mouse
666 13: 1 XT-PIC fpu
667 14: 1232265 XT-PIC ide0
668 15: 7 XT-PIC ide1
669 NMI: 0
670
671In 2.4.* a couple of lines where added to this file LOC & ERR (this time is the
672output of a SMP machine):
673
674 > cat /proc/interrupts
675
676 CPU0 CPU1
677 0: 1243498 1214548 IO-APIC-edge timer
678 1: 8949 8958 IO-APIC-edge keyboard
679 2: 0 0 XT-PIC cascade
680 5: 11286 10161 IO-APIC-edge soundblaster
681 8: 1 0 IO-APIC-edge rtc
682 9: 27422 27407 IO-APIC-edge 3c503
683 12: 113645 113873 IO-APIC-edge PS/2 Mouse
684 13: 0 0 XT-PIC fpu
685 14: 22491 24012 IO-APIC-edge ide0
686 15: 2183 2415 IO-APIC-edge ide1
687 17: 30564 30414 IO-APIC-level eth0
688 18: 177 164 IO-APIC-level bttv
689 NMI: 2457961 2457959
690 LOC: 2457882 2457881
691 ERR: 2155
692
693NMI is incremented in this case because every timer interrupt generates a NMI
694(Non Maskable Interrupt) which is used by the NMI Watchdog to detect lockups.
695
696LOC is the local interrupt counter of the internal APIC of every CPU.
697
698ERR is incremented in the case of errors in the IO-APIC bus (the bus that
699connects the CPUs in a SMP system. This means that an error has been detected,
700the IO-APIC automatically retry the transmission, so it should not be a big
701problem, but you should read the SMP-FAQ.
702
Joe Korty38e760a2007-10-17 18:04:40 +0200703In 2.6.2* /proc/interrupts was expanded again. This time the goal was for
704/proc/interrupts to display every IRQ vector in use by the system, not
705just those considered 'most important'. The new vectors are:
706
707 THR -- interrupt raised when a machine check threshold counter
708 (typically counting ECC corrected errors of memory or cache) exceeds
709 a configurable threshold. Only available on some systems.
710
711 TRM -- a thermal event interrupt occurs when a temperature threshold
712 has been exceeded for the CPU. This interrupt may also be generated
713 when the temperature drops back to normal.
714
715 SPU -- a spurious interrupt is some interrupt that was raised then lowered
716 by some IO device before it could be fully processed by the APIC. Hence
717 the APIC sees the interrupt but does not know what device it came from.
718 For this case the APIC will generate the interrupt with a IRQ vector
719 of 0xff. This might also be generated by chipset bugs.
720
721 RES, CAL, TLB -- rescheduling, call and TLB flush interrupts are
722 sent from one CPU to another per the needs of the OS. Typically,
723 their statistics are used by kernel developers and interested users to
Matt LaPlante19f59462009-04-27 15:06:31 +0200724 determine the occurrence of interrupts of the given type.
Joe Korty38e760a2007-10-17 18:04:40 +0200725
Lucas De Marchi25985ed2011-03-30 22:57:33 -0300726The above IRQ vectors are displayed only when relevant. For example,
Joe Korty38e760a2007-10-17 18:04:40 +0200727the threshold vector does not exist on x86_64 platforms. Others are
728suppressed when the system is a uniprocessor. As of this writing, only
729i386 and x86_64 platforms support the new IRQ vector displays.
730
731Of some interest is the introduction of the /proc/irq directory to 2.4.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700732It could be used to set IRQ to CPU affinity, this means that you can "hook" an
733IRQ to only one CPU, or to exclude a CPU of handling IRQs. The contents of the
Max Krasnyansky18404752008-05-29 11:02:52 -0700734irq subdir is one subdir for each IRQ, and two files; default_smp_affinity and
735prof_cpu_mask.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700736
737For example
738 > ls /proc/irq/
739 0 10 12 14 16 18 2 4 6 8 prof_cpu_mask
Max Krasnyansky18404752008-05-29 11:02:52 -0700740 1 11 13 15 17 19 3 5 7 9 default_smp_affinity
Linus Torvalds1da177e2005-04-16 15:20:36 -0700741 > ls /proc/irq/0/
742 smp_affinity
743
Max Krasnyansky18404752008-05-29 11:02:52 -0700744smp_affinity is a bitmask, in which you can specify which CPUs can handle the
745IRQ, you can set it by doing:
Linus Torvalds1da177e2005-04-16 15:20:36 -0700746
Max Krasnyansky18404752008-05-29 11:02:52 -0700747 > echo 1 > /proc/irq/10/smp_affinity
748
749This means that only the first CPU will handle the IRQ, but you can also echo
John Kacur99e9d9582016-06-17 15:05:15 +02007505 which means that only the first and third CPU can handle the IRQ.
Max Krasnyansky18404752008-05-29 11:02:52 -0700751
752The contents of each smp_affinity file is the same by default:
753
754 > cat /proc/irq/0/smp_affinity
Linus Torvalds1da177e2005-04-16 15:20:36 -0700755 ffffffff
756
Mike Travis4b0604202011-05-24 17:13:12 -0700757There is an alternate interface, smp_affinity_list which allows specifying
758a cpu range instead of a bitmask:
759
760 > cat /proc/irq/0/smp_affinity_list
761 1024-1031
762
Max Krasnyansky18404752008-05-29 11:02:52 -0700763The default_smp_affinity mask applies to all non-active IRQs, which are the
764IRQs which have not yet been allocated/activated, and hence which lack a
765/proc/irq/[0-9]* directory.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700766
Dimitri Sivanich92d6b712010-03-11 14:08:56 -0800767The node file on an SMP system shows the node to which the device using the IRQ
768reports itself as being attached. This hardware locality information does not
769include information about any possible driver locality preference.
770
Max Krasnyansky18404752008-05-29 11:02:52 -0700771prof_cpu_mask specifies which CPUs are to be profiled by the system wide
Mike Travis4b0604202011-05-24 17:13:12 -0700772profiler. Default value is ffffffff (all cpus if there are only 32 of them).
Linus Torvalds1da177e2005-04-16 15:20:36 -0700773
774The way IRQs are routed is handled by the IO-APIC, and it's Round Robin
775between all the CPUs which are allowed to handle it. As usual the kernel has
776more info than you and does a better job than you, so the defaults are the
Mike Travis4b0604202011-05-24 17:13:12 -0700777best choice for almost everyone. [Note this applies only to those IO-APIC's
778that support "Round Robin" interrupt distribution.]
Linus Torvalds1da177e2005-04-16 15:20:36 -0700779
780There are three more important subdirectories in /proc: net, scsi, and sys.
781The general rule is that the contents, or even the existence of these
782directories, depend on your kernel configuration. If SCSI is not enabled, the
783directory scsi may not exist. The same is true with the net, which is there
784only when networking support is present in the running kernel.
785
786The slabinfo file gives information about memory usage at the slab level.
787Linux uses slab pools for memory management above page level in version 2.2.
788Commonly used objects have their own slab pool (such as network buffers,
789directory cache, and so on).
790
791..............................................................................
792
793> cat /proc/buddyinfo
794
795Node 0, zone DMA 0 4 5 4 4 3 ...
796Node 0, zone Normal 1 0 0 1 101 8 ...
797Node 0, zone HighMem 2 0 0 1 1 0 ...
798
Mel Gormana1b57ac2010-03-05 13:42:15 -0800799External fragmentation is a problem under some workloads, and buddyinfo is a
Linus Torvalds1da177e2005-04-16 15:20:36 -0700800useful tool for helping diagnose these problems. Buddyinfo will give you a
801clue as to how big an area you can safely allocate, or why a previous
802allocation failed.
803
804Each column represents the number of pages of a certain order which are
805available. In this case, there are 0 chunks of 2^0*PAGE_SIZE available in
806ZONE_DMA, 4 chunks of 2^1*PAGE_SIZE in ZONE_DMA, 101 chunks of 2^4*PAGE_SIZE
807available in ZONE_NORMAL, etc...
808
Mel Gormana1b57ac2010-03-05 13:42:15 -0800809More information relevant to external fragmentation can be found in
810pagetypeinfo.
811
812> cat /proc/pagetypeinfo
813Page block order: 9
814Pages per block: 512
815
816Free pages count per migrate type at order 0 1 2 3 4 5 6 7 8 9 10
817Node 0, zone DMA, type Unmovable 0 0 0 1 1 1 1 1 1 1 0
818Node 0, zone DMA, type Reclaimable 0 0 0 0 0 0 0 0 0 0 0
819Node 0, zone DMA, type Movable 1 1 2 1 2 1 1 0 1 0 2
820Node 0, zone DMA, type Reserve 0 0 0 0 0 0 0 0 0 1 0
821Node 0, zone DMA, type Isolate 0 0 0 0 0 0 0 0 0 0 0
822Node 0, zone DMA32, type Unmovable 103 54 77 1 1 1 11 8 7 1 9
823Node 0, zone DMA32, type Reclaimable 0 0 2 1 0 0 0 0 1 0 0
824Node 0, zone DMA32, type Movable 169 152 113 91 77 54 39 13 6 1 452
825Node 0, zone DMA32, type Reserve 1 2 2 2 2 0 1 1 1 1 0
826Node 0, zone DMA32, type Isolate 0 0 0 0 0 0 0 0 0 0 0
827
828Number of blocks type Unmovable Reclaimable Movable Reserve Isolate
829Node 0, zone DMA 2 0 5 1 0
830Node 0, zone DMA32 41 6 967 2 0
831
832Fragmentation avoidance in the kernel works by grouping pages of different
833migrate types into the same contiguous regions of memory called page blocks.
834A page block is typically the size of the default hugepage size e.g. 2MB on
835X86-64. By keeping pages grouped based on their ability to move, the kernel
836can reclaim pages within a page block to satisfy a high-order allocation.
837
838The pagetypinfo begins with information on the size of a page block. It
839then gives the same type of information as buddyinfo except broken down
840by migrate-type and finishes with details on how many page blocks of each
841type exist.
842
843If min_free_kbytes has been tuned correctly (recommendations made by hugeadm
SeongJae Parkceec86ec2016-01-13 16:47:56 +0900844from libhugetlbfs https://github.com/libhugetlbfs/libhugetlbfs/), one can
Mel Gormana1b57ac2010-03-05 13:42:15 -0800845make an estimate of the likely number of huge pages that can be allocated
846at a given point in time. All the "Movable" blocks should be allocatable
847unless memory has been mlock()'d. Some of the Reclaimable blocks should
848also be allocatable although a lot of filesystem metadata may have to be
849reclaimed to achieve this.
850
Linus Torvalds1da177e2005-04-16 15:20:36 -0700851..............................................................................
852
853meminfo:
854
855Provides information about distribution and utilization of memory. This
856varies by architecture and compile options. The following is from a
85716GB PIII, which has highmem enabled. You may not have all of these fields.
858
859> cat /proc/meminfo
860
Linus Torvalds1da177e2005-04-16 15:20:36 -0700861MemTotal: 16344972 kB
862MemFree: 13634064 kB
Rik van Riel34e431b2014-01-21 15:49:05 -0800863MemAvailable: 14836172 kB
Linus Torvalds1da177e2005-04-16 15:20:36 -0700864Buffers: 3656 kB
865Cached: 1195708 kB
866SwapCached: 0 kB
867Active: 891636 kB
868Inactive: 1077224 kB
869HighTotal: 15597528 kB
870HighFree: 13629632 kB
871LowTotal: 747444 kB
872LowFree: 4432 kB
873SwapTotal: 0 kB
874SwapFree: 0 kB
875Dirty: 968 kB
876Writeback: 0 kB
Miklos Szeredib88473f2008-04-30 00:54:39 -0700877AnonPages: 861800 kB
Linus Torvalds1da177e2005-04-16 15:20:36 -0700878Mapped: 280372 kB
Rodrigo Freire0bc126d2016-01-14 15:21:58 -0800879Shmem: 644 kB
Vlastimil Babka61f94e12018-10-26 15:05:50 -0700880KReclaimable: 168048 kB
Miklos Szeredib88473f2008-04-30 00:54:39 -0700881Slab: 284364 kB
882SReclaimable: 159856 kB
883SUnreclaim: 124508 kB
884PageTables: 24448 kB
885NFS_Unstable: 0 kB
886Bounce: 0 kB
887WritebackTmp: 0 kB
Linus Torvalds1da177e2005-04-16 15:20:36 -0700888CommitLimit: 7669796 kB
889Committed_AS: 100056 kB
Linus Torvalds1da177e2005-04-16 15:20:36 -0700890VmallocTotal: 112216 kB
891VmallocUsed: 428 kB
892VmallocChunk: 111088 kB
Dennis Zhou (Facebook)7e8a6302018-08-21 21:53:58 -0700893Percpu: 62080 kB
Prashant Dhamdhere655c75a2018-07-13 22:58:06 +0530894HardwareCorrupted: 0 kB
Mel Gorman69256992012-05-29 15:06:45 -0700895AnonHugePages: 49152 kB
Kirill A. Shutemov1b5946a2016-07-26 15:26:40 -0700896ShmemHugePages: 0 kB
897ShmemPmdMapped: 0 kB
898
Linus Torvalds1da177e2005-04-16 15:20:36 -0700899
900 MemTotal: Total usable ram (i.e. physical ram minus a few reserved
901 bits and the kernel binary code)
902 MemFree: The sum of LowFree+HighFree
Rik van Riel34e431b2014-01-21 15:49:05 -0800903MemAvailable: An estimate of how much memory is available for starting new
904 applications, without swapping. Calculated from MemFree,
905 SReclaimable, the size of the file LRU lists, and the low
906 watermarks in each zone.
907 The estimate takes into account that the system needs some
908 page cache to function well, and that not all reclaimable
909 slab will be reclaimable, due to items being in use. The
910 impact of those factors will vary from system to system.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700911 Buffers: Relatively temporary storage for raw disk blocks
912 shouldn't get tremendously large (20MB or so)
913 Cached: in-memory cache for files read from the disk (the
914 pagecache). Doesn't include SwapCached
915 SwapCached: Memory that once was swapped out, is swapped back in but
916 still also is in the swapfile (if memory is needed it
917 doesn't need to be swapped out AGAIN because it is already
918 in the swapfile. This saves I/O)
919 Active: Memory that has been used more recently and usually not
920 reclaimed unless absolutely necessary.
921 Inactive: Memory which has been less recently used. It is more
922 eligible to be reclaimed for other purposes
923 HighTotal:
924 HighFree: Highmem is all memory above ~860MB of physical memory
925 Highmem areas are for use by userspace programs, or
926 for the pagecache. The kernel must use tricks to access
927 this memory, making it slower to access than lowmem.
928 LowTotal:
929 LowFree: Lowmem is memory which can be used for everything that
Matt LaPlante3f6dee92006-10-03 22:45:33 +0200930 highmem can be used for, but it is also available for the
Linus Torvalds1da177e2005-04-16 15:20:36 -0700931 kernel's use for its own data structures. Among many
932 other things, it is where everything from the Slab is
933 allocated. Bad things happen when you're out of lowmem.
934 SwapTotal: total amount of swap space available
935 SwapFree: Memory which has been evicted from RAM, and is temporarily
936 on the disk
937 Dirty: Memory which is waiting to get written back to the disk
938 Writeback: Memory which is actively being written back to the disk
Miklos Szeredib88473f2008-04-30 00:54:39 -0700939 AnonPages: Non-file backed pages mapped into userspace page tables
Prashant Dhamdhere655c75a2018-07-13 22:58:06 +0530940HardwareCorrupted: The amount of RAM/memory in KB, the kernel identifies as
941 corrupted.
Mel Gorman69256992012-05-29 15:06:45 -0700942AnonHugePages: Non-file backed huge pages mapped into userspace page tables
Linus Torvalds1da177e2005-04-16 15:20:36 -0700943 Mapped: files which have been mmaped, such as libraries
Rodrigo Freire0bc126d2016-01-14 15:21:58 -0800944 Shmem: Total memory used by shared memory (shmem) and tmpfs
Kirill A. Shutemov1b5946a2016-07-26 15:26:40 -0700945ShmemHugePages: Memory used by shared memory (shmem) and tmpfs allocated
946 with huge pages
947ShmemPmdMapped: Shared memory mapped into userspace with huge pages
Vlastimil Babka61f94e12018-10-26 15:05:50 -0700948KReclaimable: Kernel allocations that the kernel will attempt to reclaim
949 under memory pressure. Includes SReclaimable (below), and other
950 direct allocations with a shrinker.
Adrian Bunke82443c2006-01-10 00:20:30 +0100951 Slab: in-kernel data structures cache
Miklos Szeredib88473f2008-04-30 00:54:39 -0700952SReclaimable: Part of Slab, that might be reclaimed, such as caches
953 SUnreclaim: Part of Slab, that cannot be reclaimed on memory pressure
954 PageTables: amount of memory dedicated to the lowest level of page
955 tables.
956NFS_Unstable: NFS pages sent to the server, but not yet committed to stable
957 storage
958 Bounce: Memory used for block device "bounce buffers"
959WritebackTmp: Memory used by FUSE for temporary writeback buffers
Linus Torvalds1da177e2005-04-16 15:20:36 -0700960 CommitLimit: Based on the overcommit ratio ('vm.overcommit_ratio'),
961 this is the total amount of memory currently available to
962 be allocated on the system. This limit is only adhered to
963 if strict overcommit accounting is enabled (mode 2 in
964 'vm.overcommit_memory').
965 The CommitLimit is calculated with the following formula:
Petr Oros7a9e6da2014-05-22 14:04:44 +0200966 CommitLimit = ([total RAM pages] - [total huge TLB pages]) *
967 overcommit_ratio / 100 + [total swap pages]
Linus Torvalds1da177e2005-04-16 15:20:36 -0700968 For example, on a system with 1G of physical RAM and 7G
969 of swap with a `vm.overcommit_ratio` of 30 it would
970 yield a CommitLimit of 7.3G.
971 For more details, see the memory overcommit documentation
972 in vm/overcommit-accounting.
973Committed_AS: The amount of memory presently allocated on the system.
974 The committed memory is a sum of all of the memory which
975 has been allocated by processes, even if it has not been
976 "used" by them as of yet. A process which malloc()'s 1G
Minto Joseph46496022013-09-11 14:24:35 -0700977 of memory, but only touches 300M of it will show up as
978 using 1G. This 1G is memory which has been "committed" to
979 by the VM and can be used at any time by the allocating
980 application. With strict overcommit enabled on the system
981 (mode 2 in 'vm.overcommit_memory'),allocations which would
982 exceed the CommitLimit (detailed above) will not be permitted.
983 This is useful if one needs to guarantee that processes will
984 not fail due to lack of memory once that memory has been
985 successfully allocated.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700986VmallocTotal: total size of vmalloc memory area
987 VmallocUsed: amount of vmalloc area which is used
Matt LaPlante19f59462009-04-27 15:06:31 +0200988VmallocChunk: largest contiguous block of vmalloc area which is free
Dennis Zhou (Facebook)7e8a6302018-08-21 21:53:58 -0700989 Percpu: Memory allocated to the percpu allocator used to back percpu
990 allocations. This stat excludes the cost of metadata.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700991
Eric Dumazeta47a1262008-07-23 21:27:38 -0700992..............................................................................
993
994vmallocinfo:
995
996Provides information about vmalloced/vmaped areas. One line per area,
997containing the virtual address range of the area, size in bytes,
998caller information of the creator, and optional information depending
999on the kind of area :
1000
1001 pages=nr number of pages
1002 phys=addr if a physical address was specified
1003 ioremap I/O mapping (ioremap() and friends)
1004 vmalloc vmalloc() area
1005 vmap vmap()ed pages
1006 user VM_USERMAP area
1007 vpages buffer for pages pointers was vmalloced (huge area)
1008 N<node>=nr (Only on NUMA kernels)
1009 Number of pages allocated on memory node <node>
1010
1011> cat /proc/vmallocinfo
10120xffffc20000000000-0xffffc20000201000 2101248 alloc_large_system_hash+0x204 ...
1013 /0x2c0 pages=512 vmalloc N0=128 N1=128 N2=128 N3=128
10140xffffc20000201000-0xffffc20000302000 1052672 alloc_large_system_hash+0x204 ...
1015 /0x2c0 pages=256 vmalloc N0=64 N1=64 N2=64 N3=64
10160xffffc20000302000-0xffffc20000304000 8192 acpi_tb_verify_table+0x21/0x4f...
1017 phys=7fee8000 ioremap
10180xffffc20000304000-0xffffc20000307000 12288 acpi_tb_verify_table+0x21/0x4f...
1019 phys=7fee7000 ioremap
10200xffffc2000031d000-0xffffc2000031f000 8192 init_vdso_vars+0x112/0x210
10210xffffc2000031f000-0xffffc2000032b000 49152 cramfs_uncompress_init+0x2e ...
1022 /0x80 pages=11 vmalloc N0=3 N1=3 N2=2 N3=3
10230xffffc2000033a000-0xffffc2000033d000 12288 sys_swapon+0x640/0xac0 ...
1024 pages=2 vmalloc N1=2
10250xffffc20000347000-0xffffc2000034c000 20480 xt_alloc_table_info+0xfe ...
1026 /0x130 [x_tables] pages=4 vmalloc N0=4
10270xffffffffa0000000-0xffffffffa000f000 61440 sys_init_module+0xc27/0x1d00 ...
1028 pages=14 vmalloc N2=14
10290xffffffffa000f000-0xffffffffa0014000 20480 sys_init_module+0xc27/0x1d00 ...
1030 pages=4 vmalloc N1=4
10310xffffffffa0014000-0xffffffffa0017000 12288 sys_init_module+0xc27/0x1d00 ...
1032 pages=2 vmalloc N1=2
10330xffffffffa0017000-0xffffffffa0022000 45056 sys_init_module+0xc27/0x1d00 ...
1034 pages=10 vmalloc N0=10
Linus Torvalds1da177e2005-04-16 15:20:36 -07001035
Keika Kobayashid3d64df2009-06-17 16:25:55 -07001036..............................................................................
1037
1038softirqs:
1039
1040Provides counts of softirq handlers serviced since boot time, for each cpu.
1041
1042> cat /proc/softirqs
1043 CPU0 CPU1 CPU2 CPU3
1044 HI: 0 0 0 0
1045 TIMER: 27166 27120 27097 27034
1046 NET_TX: 0 0 0 17
1047 NET_RX: 42 0 0 39
1048 BLOCK: 0 0 107 1121
1049 TASKLET: 0 0 0 290
1050 SCHED: 27035 26983 26971 26746
1051 HRTIMER: 0 0 0 0
Shaohua Li09223372011-06-14 13:26:25 +08001052 RCU: 1678 1769 2178 2250
Keika Kobayashid3d64df2009-06-17 16:25:55 -07001053
1054
Linus Torvalds1da177e2005-04-16 15:20:36 -070010551.3 IDE devices in /proc/ide
1056----------------------------
1057
1058The subdirectory /proc/ide contains information about all IDE devices of which
1059the kernel is aware. There is one subdirectory for each IDE controller, the
1060file drivers and a link for each IDE device, pointing to the device directory
1061in the controller specific subtree.
1062
1063The file drivers contains general information about the drivers used for the
1064IDE devices:
1065
1066 > cat /proc/ide/drivers
1067 ide-cdrom version 4.53
1068 ide-disk version 1.08
1069
1070More detailed information can be found in the controller specific
1071subdirectories. These are named ide0, ide1 and so on. Each of these
Stefani Seibold349888e2009-06-17 16:26:01 -07001072directories contains the files shown in table 1-6.
Linus Torvalds1da177e2005-04-16 15:20:36 -07001073
1074
Stefani Seibold349888e2009-06-17 16:26:01 -07001075Table 1-6: IDE controller info in /proc/ide/ide?
Linus Torvalds1da177e2005-04-16 15:20:36 -07001076..............................................................................
1077 File Content
1078 channel IDE channel (0 or 1)
1079 config Configuration (only for PCI/IDE bridge)
1080 mate Mate name
1081 model Type/Chipset of IDE controller
1082..............................................................................
1083
1084Each device connected to a controller has a separate subdirectory in the
Stefani Seibold349888e2009-06-17 16:26:01 -07001085controllers directory. The files listed in table 1-7 are contained in these
Linus Torvalds1da177e2005-04-16 15:20:36 -07001086directories.
1087
1088
Stefani Seibold349888e2009-06-17 16:26:01 -07001089Table 1-7: IDE device information
Linus Torvalds1da177e2005-04-16 15:20:36 -07001090..............................................................................
1091 File Content
1092 cache The cache
1093 capacity Capacity of the medium (in 512Byte blocks)
1094 driver driver and version
1095 geometry physical and logical geometry
1096 identify device identify block
1097 media media type
1098 model device identifier
1099 settings device setup
1100 smart_thresholds IDE disk management thresholds
1101 smart_values IDE disk management values
1102..............................................................................
1103
1104The most interesting file is settings. This file contains a nice overview of
1105the drive parameters:
1106
1107 # cat /proc/ide/ide0/hda/settings
1108 name value min max mode
1109 ---- ----- --- --- ----
1110 bios_cyl 526 0 65535 rw
1111 bios_head 255 0 255 rw
1112 bios_sect 63 0 63 rw
1113 breada_readahead 4 0 127 rw
1114 bswap 0 0 1 r
1115 file_readahead 72 0 2097151 rw
1116 io_32bit 0 0 3 rw
1117 keepsettings 0 0 1 rw
1118 max_kb_per_request 122 1 127 rw
1119 multcount 0 0 8 rw
1120 nice1 1 0 1 rw
1121 nowerr 0 0 1 rw
1122 pio_mode write-only 0 255 w
1123 slow 0 0 1 rw
1124 unmaskirq 0 0 1 rw
1125 using_dma 0 0 1 rw
1126
1127
11281.4 Networking info in /proc/net
1129--------------------------------
1130
Stefani Seibold349888e2009-06-17 16:26:01 -07001131The subdirectory /proc/net follows the usual pattern. Table 1-8 shows the
Linus Torvalds1da177e2005-04-16 15:20:36 -07001132additional values you get for IP version 6 if you configure the kernel to
Stefani Seibold349888e2009-06-17 16:26:01 -07001133support this. Table 1-9 lists the files and their meaning.
Linus Torvalds1da177e2005-04-16 15:20:36 -07001134
1135
Stefani Seibold349888e2009-06-17 16:26:01 -07001136Table 1-8: IPv6 info in /proc/net
Linus Torvalds1da177e2005-04-16 15:20:36 -07001137..............................................................................
1138 File Content
1139 udp6 UDP sockets (IPv6)
1140 tcp6 TCP sockets (IPv6)
1141 raw6 Raw device statistics (IPv6)
1142 igmp6 IP multicast addresses, which this host joined (IPv6)
1143 if_inet6 List of IPv6 interface addresses
1144 ipv6_route Kernel routing table for IPv6
1145 rt6_stats Global IPv6 routing tables statistics
1146 sockstat6 Socket statistics (IPv6)
1147 snmp6 Snmp data (IPv6)
1148..............................................................................
1149
1150
Stefani Seibold349888e2009-06-17 16:26:01 -07001151Table 1-9: Network info in /proc/net
Linus Torvalds1da177e2005-04-16 15:20:36 -07001152..............................................................................
1153 File Content
1154 arp Kernel ARP table
1155 dev network devices with statistics
1156 dev_mcast the Layer2 multicast groups a device is listening too
1157 (interface index, label, number of references, number of bound
1158 addresses).
1159 dev_stat network device status
1160 ip_fwchains Firewall chain linkage
1161 ip_fwnames Firewall chain names
1162 ip_masq Directory containing the masquerading tables
1163 ip_masquerade Major masquerading table
1164 netstat Network statistics
1165 raw raw device statistics
1166 route Kernel routing table
1167 rpc Directory containing rpc info
1168 rt_cache Routing cache
1169 snmp SNMP data
1170 sockstat Socket statistics
1171 tcp TCP sockets
Linus Torvalds1da177e2005-04-16 15:20:36 -07001172 udp UDP sockets
1173 unix UNIX domain sockets
1174 wireless Wireless interface data (Wavelan etc)
1175 igmp IP multicast addresses, which this host joined
1176 psched Global packet scheduler parameters.
1177 netlink List of PF_NETLINK sockets
1178 ip_mr_vifs List of multicast virtual interfaces
1179 ip_mr_cache List of multicast routing cache
1180..............................................................................
1181
1182You can use this information to see which network devices are available in
1183your system and how much traffic was routed over those devices:
1184
1185 > cat /proc/net/dev
1186 Inter-|Receive |[...
1187 face |bytes packets errs drop fifo frame compressed multicast|[...
1188 lo: 908188 5596 0 0 0 0 0 0 [...
1189 ppp0:15475140 20721 410 0 0 410 0 0 [...
1190 eth0: 614530 7085 0 0 0 0 0 1 [...
1191
1192 ...] Transmit
1193 ...] bytes packets errs drop fifo colls carrier compressed
1194 ...] 908188 5596 0 0 0 0 0 0
1195 ...] 1375103 17405 0 0 0 0 0 0
1196 ...] 1703981 5535 0 0 0 3 0 0
1197
Francis Galieguea33f3222010-04-23 00:08:02 +02001198In addition, each Channel Bond interface has its own directory. For
Linus Torvalds1da177e2005-04-16 15:20:36 -07001199example, the bond0 device will have a directory called /proc/net/bond0/.
1200It will contain information that is specific to that bond, such as the
1201current slaves of the bond, the link status of the slaves, and how
1202many times the slaves link has failed.
1203
12041.5 SCSI info
1205-------------
1206
1207If you have a SCSI host adapter in your system, you'll find a subdirectory
1208named after the driver for this adapter in /proc/scsi. You'll also see a list
1209of all recognized SCSI devices in /proc/scsi:
1210
1211 >cat /proc/scsi/scsi
1212 Attached devices:
1213 Host: scsi0 Channel: 00 Id: 00 Lun: 00
1214 Vendor: IBM Model: DGHS09U Rev: 03E0
1215 Type: Direct-Access ANSI SCSI revision: 03
1216 Host: scsi0 Channel: 00 Id: 06 Lun: 00
1217 Vendor: PIONEER Model: CD-ROM DR-U06S Rev: 1.04
1218 Type: CD-ROM ANSI SCSI revision: 02
1219
1220
1221The directory named after the driver has one file for each adapter found in
1222the system. These files contain information about the controller, including
1223the used IRQ and the IO address range. The amount of information shown is
1224dependent on the adapter you use. The example shows the output for an Adaptec
1225AHA-2940 SCSI adapter:
1226
1227 > cat /proc/scsi/aic7xxx/0
1228
1229 Adaptec AIC7xxx driver version: 5.1.19/3.2.4
1230 Compile Options:
1231 TCQ Enabled By Default : Disabled
1232 AIC7XXX_PROC_STATS : Disabled
1233 AIC7XXX_RESET_DELAY : 5
1234 Adapter Configuration:
1235 SCSI Adapter: Adaptec AHA-294X Ultra SCSI host adapter
1236 Ultra Wide Controller
1237 PCI MMAPed I/O Base: 0xeb001000
1238 Adapter SEEPROM Config: SEEPROM found and used.
1239 Adaptec SCSI BIOS: Enabled
1240 IRQ: 10
1241 SCBs: Active 0, Max Active 2,
1242 Allocated 15, HW 16, Page 255
1243 Interrupts: 160328
1244 BIOS Control Word: 0x18b6
1245 Adapter Control Word: 0x005b
1246 Extended Translation: Enabled
1247 Disconnect Enable Flags: 0xffff
1248 Ultra Enable Flags: 0x0001
1249 Tag Queue Enable Flags: 0x0000
1250 Ordered Queue Tag Flags: 0x0000
1251 Default Tag Queue Depth: 8
1252 Tagged Queue By Device array for aic7xxx host instance 0:
1253 {255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255}
1254 Actual queue depth per device for aic7xxx host instance 0:
1255 {1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1}
1256 Statistics:
1257 (scsi0:0:0:0)
1258 Device using Wide/Sync transfers at 40.0 MByte/sec, offset 8
1259 Transinfo settings: current(12/8/1/0), goal(12/8/1/0), user(12/15/1/0)
1260 Total transfers 160151 (74577 reads and 85574 writes)
1261 (scsi0:0:6:0)
1262 Device using Narrow/Sync transfers at 5.0 MByte/sec, offset 15
1263 Transinfo settings: current(50/15/0/0), goal(50/15/0/0), user(50/15/0/0)
1264 Total transfers 0 (0 reads and 0 writes)
1265
1266
12671.6 Parallel port info in /proc/parport
1268---------------------------------------
1269
1270The directory /proc/parport contains information about the parallel ports of
1271your system. It has one subdirectory for each port, named after the port
1272number (0,1,2,...).
1273
Stefani Seibold349888e2009-06-17 16:26:01 -07001274These directories contain the four files shown in Table 1-10.
Linus Torvalds1da177e2005-04-16 15:20:36 -07001275
1276
Stefani Seibold349888e2009-06-17 16:26:01 -07001277Table 1-10: Files in /proc/parport
Linus Torvalds1da177e2005-04-16 15:20:36 -07001278..............................................................................
1279 File Content
1280 autoprobe Any IEEE-1284 device ID information that has been acquired.
1281 devices list of the device drivers using that port. A + will appear by the
1282 name of the device currently using the port (it might not appear
1283 against any).
1284 hardware Parallel port's base address, IRQ line and DMA channel.
1285 irq IRQ that parport is using for that port. This is in a separate
1286 file to allow you to alter it by writing a new value in (IRQ
1287 number or none).
1288..............................................................................
1289
12901.7 TTY info in /proc/tty
1291-------------------------
1292
1293Information about the available and actually used tty's can be found in the
1294directory /proc/tty.You'll find entries for drivers and line disciplines in
Stefani Seibold349888e2009-06-17 16:26:01 -07001295this directory, as shown in Table 1-11.
Linus Torvalds1da177e2005-04-16 15:20:36 -07001296
1297
Stefani Seibold349888e2009-06-17 16:26:01 -07001298Table 1-11: Files in /proc/tty
Linus Torvalds1da177e2005-04-16 15:20:36 -07001299..............................................................................
1300 File Content
1301 drivers list of drivers and their usage
1302 ldiscs registered line disciplines
1303 driver/serial usage statistic and status of single tty lines
1304..............................................................................
1305
1306To see which tty's are currently in use, you can simply look into the file
1307/proc/tty/drivers:
1308
1309 > cat /proc/tty/drivers
1310 pty_slave /dev/pts 136 0-255 pty:slave
1311 pty_master /dev/ptm 128 0-255 pty:master
1312 pty_slave /dev/ttyp 3 0-255 pty:slave
1313 pty_master /dev/pty 2 0-255 pty:master
1314 serial /dev/cua 5 64-67 serial:callout
1315 serial /dev/ttyS 4 64-67 serial
1316 /dev/tty0 /dev/tty0 4 0 system:vtmaster
1317 /dev/ptmx /dev/ptmx 5 2 system
1318 /dev/console /dev/console 5 1 system:console
1319 /dev/tty /dev/tty 5 0 system:/dev/tty
1320 unknown /dev/tty 4 1-63 console
1321
1322
13231.8 Miscellaneous kernel statistics in /proc/stat
1324-------------------------------------------------
1325
1326Various pieces of information about kernel activity are available in the
1327/proc/stat file. All of the numbers reported in this file are aggregates
1328since the system first booted. For a quick look, simply cat the file:
1329
1330 > cat /proc/stat
Tobias Klauserc8a329c2015-03-30 15:49:26 +02001331 cpu 2255 34 2290 22625563 6290 127 456 0 0 0
1332 cpu0 1132 34 1441 11311718 3675 127 438 0 0 0
1333 cpu1 1123 0 849 11313845 2614 0 18 0 0 0
Linus Torvalds1da177e2005-04-16 15:20:36 -07001334 intr 114930548 113199788 3 0 5 263 0 4 [... lots more numbers ...]
1335 ctxt 1990473
1336 btime 1062191376
1337 processes 2915
1338 procs_running 1
1339 procs_blocked 0
Keika Kobayashid3d64df2009-06-17 16:25:55 -07001340 softirq 183433 0 21755 12 39 1137 231 21459 2263
Linus Torvalds1da177e2005-04-16 15:20:36 -07001341
1342The very first "cpu" line aggregates the numbers in all of the other "cpuN"
1343lines. These numbers identify the amount of time the CPU has spent performing
1344different kinds of work. Time units are in USER_HZ (typically hundredths of a
1345second). The meanings of the columns are as follows, from left to right:
1346
1347- user: normal processes executing in user mode
1348- nice: niced processes executing in user mode
1349- system: processes executing in kernel mode
1350- idle: twiddling thumbs
Chao Fan9c240d72016-10-26 10:41:28 +08001351- iowait: In a word, iowait stands for waiting for I/O to complete. But there
1352 are several problems:
1353 1. Cpu will not wait for I/O to complete, iowait is the time that a task is
1354 waiting for I/O to complete. When cpu goes into idle state for
1355 outstanding task io, another task will be scheduled on this CPU.
1356 2. In a multi-core CPU, the task waiting for I/O to complete is not running
1357 on any CPU, so the iowait of each CPU is difficult to calculate.
1358 3. The value of iowait field in /proc/stat will decrease in certain
1359 conditions.
1360 So, the iowait is not reliable by reading from /proc/stat.
Linus Torvalds1da177e2005-04-16 15:20:36 -07001361- irq: servicing interrupts
1362- softirq: servicing softirqs
Leonardo Chiquittob68f2c3a2007-10-20 03:03:38 +02001363- steal: involuntary wait
Ryota Ozakice0e7b22009-10-24 01:20:10 +09001364- guest: running a normal guest
1365- guest_nice: running a niced guest
Linus Torvalds1da177e2005-04-16 15:20:36 -07001366
1367The "intr" line gives counts of interrupts serviced since boot time, for each
1368of the possible system interrupts. The first column is the total of all
Jan Moskyto Matejka3568a1d2014-05-15 13:55:34 -07001369interrupts serviced including unnumbered architecture specific interrupts;
1370each subsequent column is the total for that particular numbered interrupt.
1371Unnumbered interrupts are not shown, only summed into the total.
Linus Torvalds1da177e2005-04-16 15:20:36 -07001372
1373The "ctxt" line gives the total number of context switches across all CPUs.
1374
1375The "btime" line gives the time at which the system booted, in seconds since
1376the Unix epoch.
1377
1378The "processes" line gives the number of processes and threads created, which
1379includes (but is not limited to) those created by calls to the fork() and
1380clone() system calls.
1381
Luis Garces-Ericee3cc2222009-12-06 18:30:44 -08001382The "procs_running" line gives the total number of threads that are
1383running or ready to run (i.e., the total number of runnable threads).
Linus Torvalds1da177e2005-04-16 15:20:36 -07001384
1385The "procs_blocked" line gives the number of processes currently blocked,
1386waiting for I/O to complete.
1387
Keika Kobayashid3d64df2009-06-17 16:25:55 -07001388The "softirq" line gives counts of softirqs serviced since boot time, for each
1389of the possible system softirqs. The first column is the total of all
1390softirqs serviced; each subsequent column is the total for that particular
1391softirq.
1392
Theodore Ts'o37515fa2008-10-09 23:21:54 -04001393
Alex Tomasc9de5602008-01-29 00:19:52 -050013941.9 Ext4 file system parameters
Maisa Roponen690b0542014-11-24 09:54:17 +02001395-------------------------------
Alex Tomasc9de5602008-01-29 00:19:52 -05001396
Theodore Ts'o37515fa2008-10-09 23:21:54 -04001397Information about mounted ext4 file systems can be found in
1398/proc/fs/ext4. Each mounted filesystem will have a directory in
1399/proc/fs/ext4 based on its device name (i.e., /proc/fs/ext4/hdc or
1400/proc/fs/ext4/dm-0). The files in each per-device directory are shown
Stefani Seibold349888e2009-06-17 16:26:01 -07001401in Table 1-12, below.
Alex Tomasc9de5602008-01-29 00:19:52 -05001402
Stefani Seibold349888e2009-06-17 16:26:01 -07001403Table 1-12: Files in /proc/fs/ext4/<devname>
Theodore Ts'o37515fa2008-10-09 23:21:54 -04001404..............................................................................
1405 File Content
1406 mb_groups details of multiblock allocator buddy cache of free blocks
Theodore Ts'o37515fa2008-10-09 23:21:54 -04001407..............................................................................
Alex Tomasc9de5602008-01-29 00:19:52 -05001408
Jiri Slaby23308ba2010-11-04 16:20:24 +010014092.0 /proc/consoles
1410------------------
1411Shows registered system console lines.
1412
1413To see which character device lines are currently used for the system console
1414/dev/console, you may simply look into the file /proc/consoles:
1415
1416 > cat /proc/consoles
1417 tty0 -WU (ECp) 4:7
1418 ttyS0 -W- (Ep) 4:64
1419
1420The columns are:
1421
1422 device name of the device
1423 operations R = can do read operations
1424 W = can do write operations
1425 U = can do unblank
1426 flags E = it is enabled
Lucas De Marchi25985ed2011-03-30 22:57:33 -03001427 C = it is preferred console
Jiri Slaby23308ba2010-11-04 16:20:24 +01001428 B = it is primary boot console
1429 p = it is used for printk buffer
1430 b = it is not a TTY but a Braille device
1431 a = it is safe to use when cpu is offline
1432 major:minor major and minor number of the device separated by a colon
Linus Torvalds1da177e2005-04-16 15:20:36 -07001433
1434------------------------------------------------------------------------------
1435Summary
1436------------------------------------------------------------------------------
1437The /proc file system serves information about the running system. It not only
1438allows access to process data but also allows you to request the kernel status
1439by reading files in the hierarchy.
1440
1441The directory structure of /proc reflects the types of information and makes
1442it easy, if not obvious, where to look for specific data.
1443------------------------------------------------------------------------------
1444
1445------------------------------------------------------------------------------
1446CHAPTER 2: MODIFYING SYSTEM PARAMETERS
1447------------------------------------------------------------------------------
1448
1449------------------------------------------------------------------------------
1450In This Chapter
1451------------------------------------------------------------------------------
1452* Modifying kernel parameters by writing into files found in /proc/sys
1453* Exploring the files which modify certain parameters
1454* Review of the /proc/sys file tree
1455------------------------------------------------------------------------------
1456
1457
1458A very interesting part of /proc is the directory /proc/sys. This is not only
1459a source of information, it also allows you to change parameters within the
1460kernel. Be very careful when attempting this. You can optimize your system,
1461but you can also cause it to crash. Never alter kernel parameters on a
1462production system. Set up a development machine and test to make sure that
1463everything works the way you want it to. You may have no alternative but to
1464reboot the machine once an error has been made.
1465
1466To change a value, simply echo the new value into the file. An example is
1467given below in the section on the file system data. You need to be root to do
1468this. You can create your own boot script to perform this every time your
1469system boots.
1470
1471The files in /proc/sys can be used to fine tune and monitor miscellaneous and
1472general things in the operation of the Linux kernel. Since some of the files
1473can inadvertently disrupt your system, it is advisable to read both
1474documentation and source before actually making adjustments. In any case, be
1475very careful when writing to any of these files. The entries in /proc may
1476change slightly between the 2.1.* and the 2.2 kernel, so if there is any doubt
1477review the kernel documentation in the directory /usr/src/linux/Documentation.
1478This chapter is heavily based on the documentation included in the pre 2.2
1479kernels, and became part of it in version 2.2.1 of the Linux kernel.
1480
Paul Bolle395cf962011-08-15 02:02:26 +02001481Please see: Documentation/sysctl/ directory for descriptions of these
Peter W Morrealedb0fb182009-01-15 13:50:42 -08001482entries.
Andrew Morton9d0243b2006-01-08 01:00:39 -08001483
Shen Feng760df932009-04-02 16:57:20 -07001484------------------------------------------------------------------------------
1485Summary
1486------------------------------------------------------------------------------
1487Certain aspects of kernel behavior can be modified at runtime, without the
1488need to recompile the kernel, or even to reboot the system. The files in the
1489/proc/sys tree can not only be read, but also modified. You can use the echo
1490command to write value into these files, thereby changing the default settings
1491of the kernel.
1492------------------------------------------------------------------------------
Andrew Morton9d0243b2006-01-08 01:00:39 -08001493
Shen Feng760df932009-04-02 16:57:20 -07001494------------------------------------------------------------------------------
1495CHAPTER 3: PER-PROCESS PARAMETERS
1496------------------------------------------------------------------------------
Linus Torvalds1da177e2005-04-16 15:20:36 -07001497
David Rientjesfa0cbbf2012-11-12 17:53:04 -080014983.1 /proc/<pid>/oom_adj & /proc/<pid>/oom_score_adj- Adjust the oom-killer score
David Rientjesa63d83f2010-08-09 17:19:46 -07001499--------------------------------------------------------------------------------
Jan-Frode Myklebustd7ff0db2006-09-29 01:59:45 -07001500
David Rientjesfa0cbbf2012-11-12 17:53:04 -08001501These file can be used to adjust the badness heuristic used to select which
David Rientjesa63d83f2010-08-09 17:19:46 -07001502process gets killed in out of memory conditions.
Jan-Frode Myklebustd7ff0db2006-09-29 01:59:45 -07001503
David Rientjesa63d83f2010-08-09 17:19:46 -07001504The badness heuristic assigns a value to each candidate task ranging from 0
1505(never kill) to 1000 (always kill) to determine which process is targeted. The
1506units are roughly a proportion along that range of allowed memory the process
1507may allocate from based on an estimation of its current memory and swap use.
1508For example, if a task is using all allowed memory, its badness score will be
15091000. If it is using half of its allowed memory, its score will be 500.
Evgeniy Polyakov9e9e3cb2009-01-29 14:25:09 -08001510
David Rientjes778c14a2014-01-30 15:46:11 -08001511There is an additional factor included in the badness score: the current memory
1512and swap usage is discounted by 3% for root processes.
Evgeniy Polyakov9e9e3cb2009-01-29 14:25:09 -08001513
David Rientjesa63d83f2010-08-09 17:19:46 -07001514The amount of "allowed" memory depends on the context in which the oom killer
1515was called. If it is due to the memory assigned to the allocating task's cpuset
1516being exhausted, the allowed memory represents the set of mems assigned to that
1517cpuset. If it is due to a mempolicy's node(s) being exhausted, the allowed
1518memory represents the set of mempolicy nodes. If it is due to a memory
1519limit (or swap limit) being reached, the allowed memory is that configured
1520limit. Finally, if it is due to the entire system being out of memory, the
1521allowed memory represents all allocatable resources.
Evgeniy Polyakov9e9e3cb2009-01-29 14:25:09 -08001522
David Rientjesa63d83f2010-08-09 17:19:46 -07001523The value of /proc/<pid>/oom_score_adj is added to the badness score before it
1524is used to determine which task to kill. Acceptable values range from -1000
1525(OOM_SCORE_ADJ_MIN) to +1000 (OOM_SCORE_ADJ_MAX). This allows userspace to
1526polarize the preference for oom killing either by always preferring a certain
1527task or completely disabling it. The lowest possible value, -1000, is
1528equivalent to disabling oom killing entirely for that task since it will always
1529report a badness score of 0.
Evgeniy Polyakov9e9e3cb2009-01-29 14:25:09 -08001530
David Rientjesa63d83f2010-08-09 17:19:46 -07001531Consequently, it is very simple for userspace to define the amount of memory to
1532consider for each task. Setting a /proc/<pid>/oom_score_adj value of +500, for
1533example, is roughly equivalent to allowing the remainder of tasks sharing the
1534same system, cpuset, mempolicy, or memory controller resources to use at least
153550% more memory. A value of -500, on the other hand, would be roughly
1536equivalent to discounting 50% of the task's allowed memory from being considered
1537as scoring against the task.
1538
David Rientjesfa0cbbf2012-11-12 17:53:04 -08001539For backwards compatibility with previous kernels, /proc/<pid>/oom_adj may also
1540be used to tune the badness score. Its acceptable values range from -16
1541(OOM_ADJUST_MIN) to +15 (OOM_ADJUST_MAX) and a special value of -17
1542(OOM_DISABLE) to disable oom killing entirely for that task. Its value is
1543scaled linearly with /proc/<pid>/oom_score_adj.
1544
Mandeep Singh Bainesdabb16f632011-01-13 15:46:05 -08001545The value of /proc/<pid>/oom_score_adj may be reduced no lower than the last
1546value set by a CAP_SYS_RESOURCE process. To reduce the value any lower
1547requires CAP_SYS_RESOURCE.
1548
David Rientjesa63d83f2010-08-09 17:19:46 -07001549Caveat: when a parent task is selected, the oom killer will sacrifice any first
Lucas De Marchi25985ed2011-03-30 22:57:33 -03001550generation children with separate address spaces instead, if possible. This
David Rientjesa63d83f2010-08-09 17:19:46 -07001551avoids servers and important system daemons from being killed and loses the
1552minimal amount of work.
1553
Evgeniy Polyakov9e9e3cb2009-01-29 14:25:09 -08001554
Shen Feng760df932009-04-02 16:57:20 -070015553.2 /proc/<pid>/oom_score - Display current oom-killer score
Jan-Frode Myklebustd7ff0db2006-09-29 01:59:45 -07001556-------------------------------------------------------------
1557
Jan-Frode Myklebustd7ff0db2006-09-29 01:59:45 -07001558This file can be used to check the current score used by the oom-killer is for
David Rientjesfa0cbbf2012-11-12 17:53:04 -08001559any given <pid>. Use it together with /proc/<pid>/oom_score_adj to tune which
1560process should be killed in an out-of-memory situation.
1561
Roland Kletzingf9c99462007-03-05 00:30:54 -08001562
Shen Feng760df932009-04-02 16:57:20 -070015633.3 /proc/<pid>/io - Display the IO accounting fields
Roland Kletzingf9c99462007-03-05 00:30:54 -08001564-------------------------------------------------------
1565
1566This file contains IO statistics for each running process
1567
1568Example
1569-------
1570
1571test:/tmp # dd if=/dev/zero of=/tmp/test.dat &
1572[1] 3828
1573
1574test:/tmp # cat /proc/3828/io
1575rchar: 323934931
1576wchar: 323929600
1577syscr: 632687
1578syscw: 632675
1579read_bytes: 0
1580write_bytes: 323932160
1581cancelled_write_bytes: 0
1582
1583
1584Description
1585-----------
1586
1587rchar
1588-----
1589
1590I/O counter: chars read
1591The number of bytes which this task has caused to be read from storage. This
1592is simply the sum of bytes which this process passed to read() and pread().
1593It includes things like tty IO and it is unaffected by whether or not actual
1594physical disk IO was required (the read might have been satisfied from
1595pagecache)
1596
1597
1598wchar
1599-----
1600
1601I/O counter: chars written
1602The number of bytes which this task has caused, or shall cause to be written
1603to disk. Similar caveats apply here as with rchar.
1604
1605
1606syscr
1607-----
1608
1609I/O counter: read syscalls
1610Attempt to count the number of read I/O operations, i.e. syscalls like read()
1611and pread().
1612
1613
1614syscw
1615-----
1616
1617I/O counter: write syscalls
1618Attempt to count the number of write I/O operations, i.e. syscalls like
1619write() and pwrite().
1620
1621
1622read_bytes
1623----------
1624
1625I/O counter: bytes read
1626Attempt to count the number of bytes which this process really did cause to
1627be fetched from the storage layer. Done at the submit_bio() level, so it is
1628accurate for block-backed filesystems. <please add status regarding NFS and
1629CIFS at a later time>
1630
1631
1632write_bytes
1633-----------
1634
1635I/O counter: bytes written
1636Attempt to count the number of bytes which this process caused to be sent to
1637the storage layer. This is done at page-dirtying time.
1638
1639
1640cancelled_write_bytes
1641---------------------
1642
1643The big inaccuracy here is truncate. If a process writes 1MB to a file and
1644then deletes the file, it will in fact perform no writeout. But it will have
1645been accounted as having caused 1MB of write.
1646In other words: The number of bytes which this process caused to not happen,
1647by truncating pagecache. A task can cause "negative" IO too. If this task
1648truncates some dirty pagecache, some IO which another task has been accounted
Francis Galieguea33f3222010-04-23 00:08:02 +02001649for (in its write_bytes) will not be happening. We _could_ just subtract that
Roland Kletzingf9c99462007-03-05 00:30:54 -08001650from the truncating task's write_bytes, but there is information loss in doing
1651that.
1652
1653
1654Note
1655----
1656
1657At its current implementation state, this is a bit racy on 32-bit machines: if
1658process A reads process B's /proc/pid/io while process B is updating one of
1659those 64-bit counters, process A could see an intermediate result.
1660
1661
1662More information about this can be found within the taskstats documentation in
1663Documentation/accounting.
1664
Shen Feng760df932009-04-02 16:57:20 -070016653.4 /proc/<pid>/coredump_filter - Core dump filtering settings
Kawai, Hidehirobb901102007-07-19 01:48:31 -07001666---------------------------------------------------------------
1667When a process is dumped, all anonymous memory is written to a core file as
1668long as the size of the core file isn't limited. But sometimes we don't want
Ross Zwisler50378352015-10-05 16:33:36 -06001669to dump some memory segments, for example, huge shared memory or DAX.
1670Conversely, sometimes we want to save file-backed memory segments into a core
1671file, not only the individual files.
Kawai, Hidehirobb901102007-07-19 01:48:31 -07001672
1673/proc/<pid>/coredump_filter allows you to customize which memory segments
1674will be dumped when the <pid> process is dumped. coredump_filter is a bitmask
1675of memory types. If a bit of the bitmask is set, memory segments of the
1676corresponding memory type are dumped, otherwise they are not dumped.
1677
Ross Zwisler50378352015-10-05 16:33:36 -06001678The following 9 memory types are supported:
Kawai, Hidehirobb901102007-07-19 01:48:31 -07001679 - (bit 0) anonymous private memory
1680 - (bit 1) anonymous shared memory
1681 - (bit 2) file-backed private memory
1682 - (bit 3) file-backed shared memory
Hidehiro Kawaib261dfe2008-09-13 02:33:10 -07001683 - (bit 4) ELF header pages in file-backed private memory areas (it is
1684 effective only if the bit 2 is cleared)
KOSAKI Motohiroe575f112008-10-18 20:27:08 -07001685 - (bit 5) hugetlb private memory
1686 - (bit 6) hugetlb shared memory
Ross Zwisler50378352015-10-05 16:33:36 -06001687 - (bit 7) DAX private memory
1688 - (bit 8) DAX shared memory
Kawai, Hidehirobb901102007-07-19 01:48:31 -07001689
1690 Note that MMIO pages such as frame buffer are never dumped and vDSO pages
1691 are always dumped regardless of the bitmask status.
1692
Ross Zwisler50378352015-10-05 16:33:36 -06001693 Note that bits 0-4 don't affect hugetlb or DAX memory. hugetlb memory is
1694 only affected by bit 5-6, and DAX is only affected by bits 7-8.
KOSAKI Motohiroe575f112008-10-18 20:27:08 -07001695
Ross Zwisler50378352015-10-05 16:33:36 -06001696The default value of coredump_filter is 0x33; this means all anonymous memory
1697segments, ELF header pages and hugetlb private memory are dumped.
Kawai, Hidehirobb901102007-07-19 01:48:31 -07001698
1699If you don't want to dump all shared memory segments attached to pid 1234,
Ross Zwisler50378352015-10-05 16:33:36 -06001700write 0x31 to the process's proc file.
Kawai, Hidehirobb901102007-07-19 01:48:31 -07001701
Ross Zwisler50378352015-10-05 16:33:36 -06001702 $ echo 0x31 > /proc/1234/coredump_filter
Kawai, Hidehirobb901102007-07-19 01:48:31 -07001703
1704When a new process is created, the process inherits the bitmask status from its
1705parent. It is useful to set up coredump_filter before the program runs.
1706For example:
1707
1708 $ echo 0x7 > /proc/self/coredump_filter
1709 $ ./some_program
1710
Shen Feng760df932009-04-02 16:57:20 -070017113.5 /proc/<pid>/mountinfo - Information about mounts
Ram Pai2d4d4862008-03-27 13:06:25 +01001712--------------------------------------------------------
1713
1714This file contains lines of the form:
1715
171636 35 98:0 /mnt1 /mnt2 rw,noatime master:1 - ext3 /dev/root rw,errors=continue
1717(1)(2)(3) (4) (5) (6) (7) (8) (9) (10) (11)
1718
1719(1) mount ID: unique identifier of the mount (may be reused after umount)
1720(2) parent ID: ID of parent (or of self for the top of the mount tree)
1721(3) major:minor: value of st_dev for files on filesystem
1722(4) root: root of the mount within the filesystem
1723(5) mount point: mount point relative to the process's root
1724(6) mount options: per mount options
1725(7) optional fields: zero or more fields of the form "tag[:value]"
1726(8) separator: marks the end of the optional fields
1727(9) filesystem type: name of filesystem of the form "type[.subtype]"
1728(10) mount source: filesystem specific information or "none"
1729(11) super options: per super block options
1730
1731Parsers should ignore all unrecognised optional fields. Currently the
1732possible optional fields are:
1733
1734shared:X mount is shared in peer group X
1735master:X mount is slave to peer group X
Miklos Szeredi97e7e0f2008-03-27 13:06:26 +01001736propagate_from:X mount is slave and receives propagation from peer group X (*)
Ram Pai2d4d4862008-03-27 13:06:25 +01001737unbindable mount is unbindable
1738
Miklos Szeredi97e7e0f2008-03-27 13:06:26 +01001739(*) X is the closest dominant peer group under the process's root. If
1740X is the immediate master of the mount, or if there's no dominant peer
1741group under the same root, then only the "master:X" field is present
1742and not the "propagate_from:X" field.
1743
Ram Pai2d4d4862008-03-27 13:06:25 +01001744For more information on mount propagation see:
1745
1746 Documentation/filesystems/sharedsubtree.txt
1747
john stultz4614a696b2009-12-14 18:00:05 -08001748
17493.6 /proc/<pid>/comm & /proc/<pid>/task/<tid>/comm
1750--------------------------------------------------------
1751These files provide a method to access a tasks comm value. It also allows for
1752a task to set its own or one of its thread siblings comm value. The comm value
1753is limited in size compared to the cmdline value, so writing anything longer
1754then the kernel's TASK_COMM_LEN (currently 16 chars) will result in a truncated
1755comm value.
Vasiliy Kulikov04996802012-01-10 15:11:31 -08001756
1757
Cyrill Gorcunov818411612012-05-31 16:26:43 -070017583.7 /proc/<pid>/task/<tid>/children - Information about task children
1759-------------------------------------------------------------------------
1760This file provides a fast way to retrieve first level children pids
1761of a task pointed by <pid>/<tid> pair. The format is a space separated
1762stream of pids.
1763
1764Note the "first level" here -- if a child has own children they will
1765not be listed here, one needs to read /proc/<children-pid>/task/<tid>/children
1766to obtain the descendants.
1767
1768Since this interface is intended to be fast and cheap it doesn't
1769guarantee to provide precise results and some children might be
1770skipped, especially if they've exited right after we printed their
1771pids, so one need to either stop or freeze processes being inspected
1772if precise results are needed.
1773
1774
Andrey Vagin49d063c2014-04-07 15:38:34 -070017753.8 /proc/<pid>/fdinfo/<fd> - Information about opened file
Cyrill Gorcunovf1d8c162012-12-17 16:05:14 -08001776---------------------------------------------------------------
1777This file provides information associated with an opened file. The regular
Andrey Vagin49d063c2014-04-07 15:38:34 -07001778files have at least three fields -- 'pos', 'flags' and mnt_id. The 'pos'
1779represents the current offset of the opened file in decimal form [see lseek(2)
1780for details], 'flags' denotes the octal O_xxx mask the file has been
1781created with [see open(2) for details] and 'mnt_id' represents mount ID of
1782the file system containing the opened file [see 3.5 /proc/<pid>/mountinfo
1783for details].
Cyrill Gorcunovf1d8c162012-12-17 16:05:14 -08001784
1785A typical output is
1786
1787 pos: 0
1788 flags: 0100002
Andrey Vagin49d063c2014-04-07 15:38:34 -07001789 mnt_id: 19
Cyrill Gorcunovf1d8c162012-12-17 16:05:14 -08001790
Andrey Vagin6c8c9032015-04-16 12:49:38 -07001791All locks associated with a file descriptor are shown in its fdinfo too.
1792
1793lock: 1: FLOCK ADVISORY WRITE 359 00:13:11691 0 EOF
1794
Cyrill Gorcunovf1d8c162012-12-17 16:05:14 -08001795The files such as eventfd, fsnotify, signalfd, epoll among the regular pos/flags
1796pair provide additional information particular to the objects they represent.
1797
1798 Eventfd files
1799 ~~~~~~~~~~~~~
1800 pos: 0
1801 flags: 04002
Andrey Vagin49d063c2014-04-07 15:38:34 -07001802 mnt_id: 9
Cyrill Gorcunovf1d8c162012-12-17 16:05:14 -08001803 eventfd-count: 5a
1804
1805 where 'eventfd-count' is hex value of a counter.
1806
1807 Signalfd files
1808 ~~~~~~~~~~~~~~
1809 pos: 0
1810 flags: 04002
Andrey Vagin49d063c2014-04-07 15:38:34 -07001811 mnt_id: 9
Cyrill Gorcunovf1d8c162012-12-17 16:05:14 -08001812 sigmask: 0000000000000200
1813
1814 where 'sigmask' is hex value of the signal mask associated
1815 with a file.
1816
1817 Epoll files
1818 ~~~~~~~~~~~
1819 pos: 0
1820 flags: 02
Andrey Vagin49d063c2014-04-07 15:38:34 -07001821 mnt_id: 9
Cyrill Gorcunov77493f02017-07-12 14:34:25 -07001822 tfd: 5 events: 1d data: ffffffffffffffff pos:0 ino:61af sdev:7
Cyrill Gorcunovf1d8c162012-12-17 16:05:14 -08001823
1824 where 'tfd' is a target file descriptor number in decimal form,
1825 'events' is events mask being watched and the 'data' is data
1826 associated with a target [see epoll(7) for more details].
1827
Cyrill Gorcunov77493f02017-07-12 14:34:25 -07001828 The 'pos' is current offset of the target file in decimal form
1829 [see lseek(2)], 'ino' and 'sdev' are inode and device numbers
1830 where target file resides, all in hex format.
1831
Cyrill Gorcunovf1d8c162012-12-17 16:05:14 -08001832 Fsnotify files
1833 ~~~~~~~~~~~~~~
1834 For inotify files the format is the following
1835
1836 pos: 0
1837 flags: 02000000
1838 inotify wd:3 ino:9e7e sdev:800013 mask:800afce ignored_mask:0 fhandle-bytes:8 fhandle-type:1 f_handle:7e9e0000640d1b6d
1839
1840 where 'wd' is a watch descriptor in decimal form, ie a target file
1841 descriptor number, 'ino' and 'sdev' are inode and device where the
1842 target file resides and the 'mask' is the mask of events, all in hex
1843 form [see inotify(7) for more details].
1844
1845 If the kernel was built with exportfs support, the path to the target
1846 file is encoded as a file handle. The file handle is provided by three
1847 fields 'fhandle-bytes', 'fhandle-type' and 'f_handle', all in hex
1848 format.
1849
1850 If the kernel is built without exportfs support the file handle won't be
1851 printed out.
1852
Cyrill Gorcunove71ec592012-12-17 16:05:18 -08001853 If there is no inotify mark attached yet the 'inotify' line will be omitted.
1854
Cyrill Gorcunovf1d8c162012-12-17 16:05:14 -08001855 For fanotify files the format is
1856
1857 pos: 0
1858 flags: 02
Andrey Vagin49d063c2014-04-07 15:38:34 -07001859 mnt_id: 9
Cyrill Gorcunove71ec592012-12-17 16:05:18 -08001860 fanotify flags:10 event-flags:0
1861 fanotify mnt_id:12 mflags:40 mask:38 ignored_mask:40000003
1862 fanotify ino:4f969 sdev:800013 mflags:0 mask:3b ignored_mask:40000000 fhandle-bytes:8 fhandle-type:1 f_handle:69f90400c275b5b4
Cyrill Gorcunovf1d8c162012-12-17 16:05:14 -08001863
Cyrill Gorcunove71ec592012-12-17 16:05:18 -08001864 where fanotify 'flags' and 'event-flags' are values used in fanotify_init
1865 call, 'mnt_id' is the mount point identifier, 'mflags' is the value of
1866 flags associated with mark which are tracked separately from events
1867 mask. 'ino', 'sdev' are target inode and device, 'mask' is the events
1868 mask and 'ignored_mask' is the mask of events which are to be ignored.
1869 All in hex format. Incorporation of 'mflags', 'mask' and 'ignored_mask'
1870 does provide information about flags and mask used in fanotify_mark
1871 call [see fsnotify manpage for details].
Cyrill Gorcunovf1d8c162012-12-17 16:05:14 -08001872
Cyrill Gorcunove71ec592012-12-17 16:05:18 -08001873 While the first three lines are mandatory and always printed, the rest is
1874 optional and may be omitted if no marks created yet.
Cyrill Gorcunovf1d8c162012-12-17 16:05:14 -08001875
Cyrill Gorcunov854d06d2014-07-16 01:54:53 +04001876 Timerfd files
1877 ~~~~~~~~~~~~~
1878
1879 pos: 0
1880 flags: 02
1881 mnt_id: 9
1882 clockid: 0
1883 ticks: 0
1884 settime flags: 01
1885 it_value: (0, 49406829)
1886 it_interval: (1, 0)
1887
1888 where 'clockid' is the clock type and 'ticks' is the number of the timer expirations
1889 that have occurred [see timerfd_create(2) for details]. 'settime flags' are
1890 flags in octal form been used to setup the timer [see timerfd_settime(2) for
1891 details]. 'it_value' is remaining time until the timer exiration.
1892 'it_interval' is the interval for the timer. Note the timer might be set up
1893 with TIMER_ABSTIME option which will be shown in 'settime flags', but 'it_value'
1894 still exhibits timer's remaining time.
Cyrill Gorcunovf1d8c162012-12-17 16:05:14 -08001895
Cyrill Gorcunov740a5dd2015-02-11 15:28:31 -080018963.9 /proc/<pid>/map_files - Information about memory mapped files
1897---------------------------------------------------------------------
1898This directory contains symbolic links which represent memory mapped files
1899the process is maintaining. Example output:
1900
1901 | lr-------- 1 root root 64 Jan 27 11:24 333c600000-333c620000 -> /usr/lib64/ld-2.18.so
1902 | lr-------- 1 root root 64 Jan 27 11:24 333c81f000-333c820000 -> /usr/lib64/ld-2.18.so
1903 | lr-------- 1 root root 64 Jan 27 11:24 333c820000-333c821000 -> /usr/lib64/ld-2.18.so
1904 | ...
1905 | lr-------- 1 root root 64 Jan 27 11:24 35d0421000-35d0422000 -> /usr/lib64/libselinux.so.1
1906 | lr-------- 1 root root 64 Jan 27 11:24 400000-41a000 -> /usr/bin/ls
1907
1908The name of a link represents the virtual memory bounds of a mapping, i.e.
1909vm_area_struct::vm_start-vm_area_struct::vm_end.
1910
1911The main purpose of the map_files is to retrieve a set of memory mapped
1912files in a fast way instead of parsing /proc/<pid>/maps or
1913/proc/<pid>/smaps, both of which contain many more records. At the same
1914time one can open(2) mappings from the listings of two processes and
1915comparing their inode numbers to figure out which anonymous memory areas
1916are actually shared.
1917
John Stultz5de23d42016-03-17 14:20:54 -070019183.10 /proc/<pid>/timerslack_ns - Task timerslack value
1919---------------------------------------------------------
1920This file provides the value of the task's timerslack value in nanoseconds.
1921This value specifies a amount of time that normal timers may be deferred
1922in order to coalesce timers and avoid unnecessary wakeups.
1923
1924This allows a task's interactivity vs power consumption trade off to be
1925adjusted.
1926
1927Writing 0 to the file will set the tasks timerslack to the default value.
1928
1929Valid values are from 0 - ULLONG_MAX
1930
1931An application setting the value must have PTRACE_MODE_ATTACH_FSCREDS level
1932permissions on the task specified to change its timerslack_ns value.
1933
Josh Poimboeuf7c23b332017-02-13 19:42:41 -060019343.11 /proc/<pid>/patch_state - Livepatch patch operation state
1935-----------------------------------------------------------------
1936When CONFIG_LIVEPATCH is enabled, this file displays the value of the
1937patch state for the task.
1938
1939A value of '-1' indicates that no patch is in transition.
1940
1941A value of '0' indicates that a patch is in transition and the task is
1942unpatched. If the patch is being enabled, then the task hasn't been
1943patched yet. If the patch is being disabled, then the task has already
1944been unpatched.
1945
1946A value of '1' indicates that a patch is in transition and the task is
1947patched. If the patch is being enabled, then the task has already been
1948patched. If the patch is being disabled, then the task hasn't been
1949unpatched yet.
1950
John Stultz5de23d42016-03-17 14:20:54 -07001951
Vasiliy Kulikov04996802012-01-10 15:11:31 -08001952------------------------------------------------------------------------------
1953Configuring procfs
1954------------------------------------------------------------------------------
1955
19564.1 Mount options
1957---------------------
1958
1959The following mount options are supported:
1960
1961 hidepid= Set /proc/<pid>/ access mode.
1962 gid= Set the group authorized to learn processes information.
1963
1964hidepid=0 means classic mode - everybody may access all /proc/<pid>/ directories
1965(default).
1966
1967hidepid=1 means users may not access any /proc/<pid>/ directories but their
1968own. Sensitive files like cmdline, sched*, status are now protected against
1969other users. This makes it impossible to learn whether any user runs
1970specific program (given the program doesn't reveal itself by its behaviour).
1971As an additional bonus, as /proc/<pid>/cmdline is unaccessible for other users,
1972poorly written programs passing sensitive information via program arguments are
1973now protected against local eavesdroppers.
1974
1975hidepid=2 means hidepid=1 plus all /proc/<pid>/ will be fully invisible to other
1976users. It doesn't mean that it hides a fact whether a process with a specific
1977pid value exists (it can be learned by other means, e.g. by "kill -0 $PID"),
1978but it hides process' uid and gid, which may be learned by stat()'ing
1979/proc/<pid>/ otherwise. It greatly complicates an intruder's task of gathering
1980information about running processes, whether some daemon runs with elevated
1981privileges, whether other user runs some sensitive program, whether other users
1982run any program at all, etc.
1983
1984gid= defines a group authorized to learn processes information otherwise
1985prohibited by hidepid=. If you use some daemon like identd which needs to learn
1986information about processes information, just add identd to this group.