1 ------------------------------------------------------------------------------
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>
7 2.4.x update Jorge Nerin <comandante@zaralinux.com> November 14 2000
8 move /proc/sys Shen Feng <shen@cn.fujitsu.com> April 1 2009
9 ------------------------------------------------------------------------------
10 Version 1.3 Kernel version 2.2.12
11 Kernel version 2.4.0-test11-pre4
12 ------------------------------------------------------------------------------
13 fixes/update part 1.1 Stefani Seibold <stefani@seibold.net> June 9 2009
19 0.1 Introduction/Credits
22 1 Collecting System Information
23 1.1 Process-Specific Subdirectories
25 1.3 IDE devices in /proc/ide
26 1.4 Networking info in /proc/net
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
31 1.9 Ext4 file system parameters
33 2 Modifying System Parameters
35 3 Per-Process Parameters
36 3.1 /proc/<pid>/oom_adj & /proc/<pid>/oom_score_adj - Adjust the oom-killer
38 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
42 3.6 /proc/<pid>/comm & /proc/<pid>/task/<tid>/comm
43 3.7 /proc/<pid>/task/<tid>/children - Information about task children
44 3.8 /proc/<pid>/fdinfo/<fd> - Information about opened file
45 3.9 /proc/<pid>/map_files - Information about memory mapped files
50 ------------------------------------------------------------------------------
52 ------------------------------------------------------------------------------
54 0.1 Introduction/Credits
55 ------------------------
57 This documentation is part of a soon (or so we hope) to be released book on
58 the SuSE Linux distribution. As there is no complete documentation for the
59 /proc file system and we've used many freely available sources to write these
60 chapters, it seems only fair to give the work back to the Linux community.
61 This work is based on the 2.2.* kernel version and the upcoming 2.4.*. I'm
62 afraid it's still far from complete, but we hope it will be useful. As far as
63 we know, it is the first 'all-in-one' document about the /proc file system. It
64 is focused on the Intel x86 hardware, so if you are looking for PPC, ARM,
65 SPARC, AXP, etc., features, you probably won't find what you are looking for.
66 It also only covers IPv4 networking, not IPv6 nor other protocols - sorry. But
67 additions and patches are welcome and will be added to this document if you
70 We'd like to thank Alan Cox, Rik van Riel, and Alexey Kuznetsov and a lot of
71 other people for help compiling this documentation. We'd also like to extend a
72 special thank you to Andi Kleen for documentation, which we relied on heavily
73 to create this document, as well as the additional information he provided.
74 Thanks to everybody else who contributed source or docs to the Linux kernel
75 and helped create a great piece of software... :)
77 If you have any comments, corrections or additions, please don't hesitate to
78 contact Bodo Bauer at bb@ricochet.net. We'll be happy to add them to this
81 The latest version of this document is available online at
82 http://tldp.org/LDP/Linux-Filesystem-Hierarchy/html/proc.html
84 If the above direction does not works for you, you could try the kernel
85 mailing list at linux-kernel@vger.kernel.org and/or try to reach me at
86 comandante@zaralinux.com.
91 We don't guarantee the correctness of this document, and if you come to us
92 complaining about how you screwed up your system because of incorrect
93 documentation, we won't feel responsible...
95 ------------------------------------------------------------------------------
96 CHAPTER 1: COLLECTING SYSTEM INFORMATION
97 ------------------------------------------------------------------------------
99 ------------------------------------------------------------------------------
101 ------------------------------------------------------------------------------
102 * Investigating the properties of the pseudo file system /proc and its
103 ability to provide information on the running Linux system
104 * Examining /proc's structure
105 * Uncovering various information about the kernel and the processes running
107 ------------------------------------------------------------------------------
110 The proc file system acts as an interface to internal data structures in the
111 kernel. It can be used to obtain information about the system and to change
112 certain kernel parameters at runtime (sysctl).
114 First, we'll take a look at the read-only parts of /proc. In Chapter 2, we
115 show you how you can use /proc/sys to change settings.
117 1.1 Process-Specific Subdirectories
118 -----------------------------------
120 The directory /proc contains (among other things) one subdirectory for each
121 process running on the system, which is named after the process ID (PID).
123 The link self points to the process reading the file system. Each process
124 subdirectory has the entries listed in Table 1-1.
127 Table 1-1: Process specific entries in /proc
128 ..............................................................................
130 clear_refs Clears page referenced bits shown in smaps output
131 cmdline Command line arguments
132 cpu Current and last cpu in which it was executed (2.4)(smp)
133 cwd Link to the current working directory
134 environ Values of environment variables
135 exe Link to the executable of this process
136 fd Directory, which contains all file descriptors
137 maps Memory maps to executables and library files (2.4)
138 mem Memory held by this process
139 root Link to the root directory of this process
141 statm Process memory status information
142 status Process status in human readable form
143 wchan Present with CONFIG_KALLSYMS=y: it shows the kernel function
144 symbol the task is blocked in - or "0" if not blocked.
146 stack Report full stack trace, enable via CONFIG_STACKTRACE
147 smaps a extension based on maps, showing the memory consumption of
148 each mapping and flags associated with it
149 numa_maps an extension based on maps, showing the memory locality and
150 binding policy as well as mem usage (in pages) of each mapping.
151 ..............................................................................
153 For example, to get the status information of a process, all you have to do is
154 read the file /proc/PID/status:
156 >cat /proc/self/status
181 SigPnd: 0000000000000000
182 ShdPnd: 0000000000000000
183 SigBlk: 0000000000000000
184 SigIgn: 0000000000000000
185 SigCgt: 0000000000000000
186 CapInh: 00000000fffffeff
187 CapPrm: 0000000000000000
188 CapEff: 0000000000000000
189 CapBnd: ffffffffffffffff
191 voluntary_ctxt_switches: 0
192 nonvoluntary_ctxt_switches: 1
194 This shows you nearly the same information you would get if you viewed it with
195 the ps command. In fact, ps uses the proc file system to obtain its
196 information. But you get a more detailed view of the process by reading the
197 file /proc/PID/status. It fields are described in table 1-2.
199 The statm file contains more detailed information about the process
200 memory usage. Its seven fields are explained in Table 1-3. The stat file
201 contains details information about the process itself. Its fields are
202 explained in Table 1-4.
204 (for SMP CONFIG users)
205 For making accounting scalable, RSS related information are handled in an
206 asynchronous manner and the value may not be very precise. To see a precise
207 snapshot of a moment, you can see /proc/<pid>/smaps file and scan page table.
208 It's slow but very precise.
210 Table 1-2: Contents of the status files (as of 4.1)
211 ..............................................................................
213 Name filename of the executable
214 State state (R is running, S is sleeping, D is sleeping
215 in an uninterruptible wait, Z is zombie,
216 T is traced or stopped)
218 Ngid NUMA group ID (0 if none)
220 PPid process id of the parent process
221 TracerPid PID of process tracing this process (0 if not)
222 Uid Real, effective, saved set, and file system UIDs
223 Gid Real, effective, saved set, and file system GIDs
224 FDSize number of file descriptor slots currently allocated
225 Groups supplementary group list
226 NStgid descendant namespace thread group ID hierarchy
227 NSpid descendant namespace process ID hierarchy
228 NSpgid descendant namespace process group ID hierarchy
229 NSsid descendant namespace session ID hierarchy
230 VmPeak peak virtual memory size
231 VmSize total program size
232 VmLck locked memory size
233 VmHWM peak resident set size ("high water mark")
234 VmRSS size of memory portions
235 VmData size of data, stack, and text segments
236 VmStk size of data, stack, and text segments
237 VmExe size of text segment
238 VmLib size of shared library code
239 VmPTE size of page table entries
240 VmPMD size of second level page tables
241 VmSwap size of swap usage (the number of referred swapents)
242 HugetlbPages size of hugetlb memory portions
243 Threads number of threads
244 SigQ number of signals queued/max. number for queue
245 SigPnd bitmap of pending signals for the thread
246 ShdPnd bitmap of shared pending signals for the process
247 SigBlk bitmap of blocked signals
248 SigIgn bitmap of ignored signals
249 SigCgt bitmap of caught signals
250 CapInh bitmap of inheritable capabilities
251 CapPrm bitmap of permitted capabilities
252 CapEff bitmap of effective capabilities
253 CapBnd bitmap of capabilities bounding set
254 Seccomp seccomp mode, like prctl(PR_GET_SECCOMP, ...)
255 Cpus_allowed mask of CPUs on which this process may run
256 Cpus_allowed_list Same as previous, but in "list format"
257 Mems_allowed mask of memory nodes allowed to this process
258 Mems_allowed_list Same as previous, but in "list format"
259 voluntary_ctxt_switches number of voluntary context switches
260 nonvoluntary_ctxt_switches number of non voluntary context switches
261 ..............................................................................
263 Table 1-3: Contents of the statm files (as of 2.6.8-rc3)
264 ..............................................................................
266 size total program size (pages) (same as VmSize in status)
267 resident size of memory portions (pages) (same as VmRSS in status)
268 shared number of pages that are shared (i.e. backed by a file)
269 trs number of pages that are 'code' (not including libs; broken,
270 includes data segment)
271 lrs number of pages of library (always 0 on 2.6)
272 drs number of pages of data/stack (including libs; broken,
273 includes library text)
274 dt number of dirty pages (always 0 on 2.6)
275 ..............................................................................
278 Table 1-4: Contents of the stat files (as of 2.6.30-rc7)
279 ..............................................................................
282 tcomm filename of the executable
283 state state (R is running, S is sleeping, D is sleeping in an
284 uninterruptible wait, Z is zombie, T is traced or stopped)
285 ppid process id of the parent process
286 pgrp pgrp of the process
288 tty_nr tty the process uses
289 tty_pgrp pgrp of the tty
291 min_flt number of minor faults
292 cmin_flt number of minor faults with child's
293 maj_flt number of major faults
294 cmaj_flt number of major faults with child's
295 utime user mode jiffies
296 stime kernel mode jiffies
297 cutime user mode jiffies with child's
298 cstime kernel mode jiffies with child's
299 priority priority level
301 num_threads number of threads
302 it_real_value (obsolete, always 0)
303 start_time time the process started after system boot
304 vsize virtual memory size
305 rss resident set memory size
306 rsslim current limit in bytes on the rss
307 start_code address above which program text can run
308 end_code address below which program text can run
309 start_stack address of the start of the main process stack
310 esp current value of ESP
311 eip current value of EIP
312 pending bitmap of pending signals
313 blocked bitmap of blocked signals
314 sigign bitmap of ignored signals
315 sigcatch bitmap of caught signals
316 0 (place holder, used to be the wchan address, use /proc/PID/wchan instead)
319 exit_signal signal to send to parent thread on exit
320 task_cpu which CPU the task is scheduled on
321 rt_priority realtime priority
322 policy scheduling policy (man sched_setscheduler)
323 blkio_ticks time spent waiting for block IO
324 gtime guest time of the task in jiffies
325 cgtime guest time of the task children in jiffies
326 start_data address above which program data+bss is placed
327 end_data address below which program data+bss is placed
328 start_brk address above which program heap can be expanded with brk()
329 arg_start address above which program command line is placed
330 arg_end address below which program command line is placed
331 env_start address above which program environment is placed
332 env_end address below which program environment is placed
333 exit_code the thread's exit_code in the form reported by the waitpid system call
334 ..............................................................................
336 The /proc/PID/maps file containing the currently mapped memory regions and
337 their access permissions.
341 address perms offset dev inode pathname
343 08048000-08049000 r-xp 00000000 03:00 8312 /opt/test
344 08049000-0804a000 rw-p 00001000 03:00 8312 /opt/test
345 0804a000-0806b000 rw-p 00000000 00:00 0 [heap]
346 a7cb1000-a7cb2000 ---p 00000000 00:00 0
347 a7cb2000-a7eb2000 rw-p 00000000 00:00 0
348 a7eb2000-a7eb3000 ---p 00000000 00:00 0
349 a7eb3000-a7ed5000 rw-p 00000000 00:00 0
350 a7ed5000-a8008000 r-xp 00000000 03:00 4222 /lib/libc.so.6
351 a8008000-a800a000 r--p 00133000 03:00 4222 /lib/libc.so.6
352 a800a000-a800b000 rw-p 00135000 03:00 4222 /lib/libc.so.6
353 a800b000-a800e000 rw-p 00000000 00:00 0
354 a800e000-a8022000 r-xp 00000000 03:00 14462 /lib/libpthread.so.0
355 a8022000-a8023000 r--p 00013000 03:00 14462 /lib/libpthread.so.0
356 a8023000-a8024000 rw-p 00014000 03:00 14462 /lib/libpthread.so.0
357 a8024000-a8027000 rw-p 00000000 00:00 0
358 a8027000-a8043000 r-xp 00000000 03:00 8317 /lib/ld-linux.so.2
359 a8043000-a8044000 r--p 0001b000 03:00 8317 /lib/ld-linux.so.2
360 a8044000-a8045000 rw-p 0001c000 03:00 8317 /lib/ld-linux.so.2
361 aff35000-aff4a000 rw-p 00000000 00:00 0 [stack]
362 ffffe000-fffff000 r-xp 00000000 00:00 0 [vdso]
364 where "address" is the address space in the process that it occupies, "perms"
365 is a set of permissions:
371 p = private (copy on write)
373 "offset" is the offset into the mapping, "dev" is the device (major:minor), and
374 "inode" is the inode on that device. 0 indicates that no inode is associated
375 with the memory region, as the case would be with BSS (uninitialized data).
376 The "pathname" shows the name associated file for this mapping. If the mapping
377 is not associated with a file:
379 [heap] = the heap of the program
380 [stack] = the stack of the main process
381 [vdso] = the "virtual dynamic shared object",
382 the kernel system call handler
384 or if empty, the mapping is anonymous.
386 The /proc/PID/smaps is an extension based on maps, showing the memory
387 consumption for each of the process's mappings. For each of mappings there
388 is a series of lines such as the following:
390 08048000-080bc000 r-xp 00000000 03:02 13130 /bin/bash
402 Private_Hugetlb: 0 kB
408 VmFlags: rd ex mr mw me dw
410 the first of these lines shows the same information as is displayed for the
411 mapping in /proc/PID/maps. The remaining lines show the size of the mapping
412 (size), the amount of the mapping that is currently resident in RAM (RSS), the
413 process' proportional share of this mapping (PSS), the number of clean and
414 dirty private pages in the mapping.
416 The "proportional set size" (PSS) of a process is the count of pages it has
417 in memory, where each page is divided by the number of processes sharing it.
418 So if a process has 1000 pages all to itself, and 1000 shared with one other
419 process, its PSS will be 1500.
420 Note that even a page which is part of a MAP_SHARED mapping, but has only
421 a single pte mapped, i.e. is currently used by only one process, is accounted
422 as private and not as shared.
423 "Referenced" indicates the amount of memory currently marked as referenced or
425 "Anonymous" shows the amount of memory that does not belong to any file. Even
426 a mapping associated with a file may contain anonymous pages: when MAP_PRIVATE
427 and a page is modified, the file page is replaced by a private anonymous copy.
428 "AnonHugePages" shows the ammount of memory backed by transparent hugepage.
429 "Shared_Hugetlb" and "Private_Hugetlb" show the ammounts of memory backed by
430 hugetlbfs page which is *not* counted in "RSS" or "PSS" field for historical
431 reasons. And these are not included in {Shared,Private}_{Clean,Dirty} field.
432 "Swap" shows how much would-be-anonymous memory is also used, but out on swap.
433 "SwapPss" shows proportional swap share of this mapping.
434 "Locked" indicates whether the mapping is locked in memory or not.
436 "VmFlags" field deserves a separate description. This member represents the kernel
437 flags associated with the particular virtual memory area in two letter encoded
438 manner. The codes are the following:
447 gd - stack segment growns down
449 dw - disabled write to the mapped file
450 lo - pages are locked in memory
451 io - memory mapped I/O area
452 sr - sequential read advise provided
453 rr - random read advise provided
454 dc - do not copy area on fork
455 de - do not expand area on remapping
456 ac - area is accountable
457 nr - swap space is not reserved for the area
458 ht - area uses huge tlb pages
459 ar - architecture specific flag
460 dd - do not include area into core dump
463 hg - huge page advise flag
464 nh - no-huge page advise flag
465 mg - mergable advise flag
467 Note that there is no guarantee that every flag and associated mnemonic will
468 be present in all further kernel releases. Things get changed, the flags may
469 be vanished or the reverse -- new added. Interpretation of their meaning
470 might change in future as well. So each consumer of these flags has to
471 follow each specific kernel version for the exact semantic.
473 This file is only present if the CONFIG_MMU kernel configuration option is
476 The /proc/PID/clear_refs is used to reset the PG_Referenced and ACCESSED/YOUNG
477 bits on both physical and virtual pages associated with a process, and the
478 soft-dirty bit on pte (see Documentation/vm/soft-dirty.txt for details).
479 To clear the bits for all the pages associated with the process
480 > echo 1 > /proc/PID/clear_refs
482 To clear the bits for the anonymous pages associated with the process
483 > echo 2 > /proc/PID/clear_refs
485 To clear the bits for the file mapped pages associated with the process
486 > echo 3 > /proc/PID/clear_refs
488 To clear the soft-dirty bit
489 > echo 4 > /proc/PID/clear_refs
491 To reset the peak resident set size ("high water mark") to the process's
493 > echo 5 > /proc/PID/clear_refs
495 Any other value written to /proc/PID/clear_refs will have no effect.
497 The /proc/pid/pagemap gives the PFN, which can be used to find the pageflags
498 using /proc/kpageflags and number of times a page is mapped using
499 /proc/kpagecount. For detailed explanation, see Documentation/vm/pagemap.txt.
501 The /proc/pid/numa_maps is an extension based on maps, showing the memory
502 locality and binding policy, as well as the memory usage (in pages) of
503 each mapping. The output follows a general format where mapping details get
504 summarized separated by blank spaces, one mapping per each file line:
506 address policy mapping details
508 00400000 default file=/usr/local/bin/app mapped=1 active=0 N3=1 kernelpagesize_kB=4
509 00600000 default file=/usr/local/bin/app anon=1 dirty=1 N3=1 kernelpagesize_kB=4
510 3206000000 default file=/lib64/ld-2.12.so mapped=26 mapmax=6 N0=24 N3=2 kernelpagesize_kB=4
511 320621f000 default file=/lib64/ld-2.12.so anon=1 dirty=1 N3=1 kernelpagesize_kB=4
512 3206220000 default file=/lib64/ld-2.12.so anon=1 dirty=1 N3=1 kernelpagesize_kB=4
513 3206221000 default anon=1 dirty=1 N3=1 kernelpagesize_kB=4
514 3206800000 default file=/lib64/libc-2.12.so mapped=59 mapmax=21 active=55 N0=41 N3=18 kernelpagesize_kB=4
515 320698b000 default file=/lib64/libc-2.12.so
516 3206b8a000 default file=/lib64/libc-2.12.so anon=2 dirty=2 N3=2 kernelpagesize_kB=4
517 3206b8e000 default file=/lib64/libc-2.12.so anon=1 dirty=1 N3=1 kernelpagesize_kB=4
518 3206b8f000 default anon=3 dirty=3 active=1 N3=3 kernelpagesize_kB=4
519 7f4dc10a2000 default anon=3 dirty=3 N3=3 kernelpagesize_kB=4
520 7f4dc10b4000 default anon=2 dirty=2 active=1 N3=2 kernelpagesize_kB=4
521 7f4dc1200000 default file=/anon_hugepage\040(deleted) huge anon=1 dirty=1 N3=1 kernelpagesize_kB=2048
522 7fff335f0000 default stack anon=3 dirty=3 N3=3 kernelpagesize_kB=4
523 7fff3369d000 default mapped=1 mapmax=35 active=0 N3=1 kernelpagesize_kB=4
526 "address" is the starting address for the mapping;
527 "policy" reports the NUMA memory policy set for the mapping (see vm/numa_memory_policy.txt);
528 "mapping details" summarizes mapping data such as mapping type, page usage counters,
529 node locality page counters (N0 == node0, N1 == node1, ...) and the kernel page
530 size, in KB, that is backing the mapping up.
535 Similar to the process entries, the kernel data files give information about
536 the running kernel. The files used to obtain this information are contained in
537 /proc and are listed in Table 1-5. Not all of these will be present in your
538 system. It depends on the kernel configuration and the loaded modules, which
539 files are there, and which are missing.
541 Table 1-5: Kernel info in /proc
542 ..............................................................................
544 apm Advanced power management info
545 buddyinfo Kernel memory allocator information (see text) (2.5)
546 bus Directory containing bus specific information
547 cmdline Kernel command line
548 cpuinfo Info about the CPU
549 devices Available devices (block and character)
550 dma Used DMS channels
551 filesystems Supported filesystems
552 driver Various drivers grouped here, currently rtc (2.4)
553 execdomains Execdomains, related to security (2.4)
554 fb Frame Buffer devices (2.4)
555 fs File system parameters, currently nfs/exports (2.4)
556 ide Directory containing info about the IDE subsystem
557 interrupts Interrupt usage
558 iomem Memory map (2.4)
559 ioports I/O port usage
560 irq Masks for irq to cpu affinity (2.4)(smp?)
561 isapnp ISA PnP (Plug&Play) Info (2.4)
562 kcore Kernel core image (can be ELF or A.OUT(deprecated in 2.4))
564 ksyms Kernel symbol table
565 loadavg Load average of last 1, 5 & 15 minutes
569 modules List of loaded modules
570 mounts Mounted filesystems
571 net Networking info (see text)
572 pagetypeinfo Additional page allocator information (see text) (2.5)
573 partitions Table of partitions known to the system
574 pci Deprecated info of PCI bus (new way -> /proc/bus/pci/,
575 decoupled by lspci (2.4)
577 scsi SCSI info (see text)
578 slabinfo Slab pool info
579 softirqs softirq usage
580 stat Overall statistics
581 swaps Swap space utilization
583 sysvipc Info of SysVIPC Resources (msg, sem, shm) (2.4)
584 tty Info of tty drivers
585 uptime Wall clock since boot, combined idle time of all cpus
586 version Kernel version
587 video bttv info of video resources (2.4)
588 vmallocinfo Show vmalloced areas
589 ..............................................................................
591 You can, for example, check which interrupts are currently in use and what
592 they are used for by looking in the file /proc/interrupts:
594 > cat /proc/interrupts
596 0: 8728810 XT-PIC timer
597 1: 895 XT-PIC keyboard
599 3: 531695 XT-PIC aha152x
600 4: 2014133 XT-PIC serial
601 5: 44401 XT-PIC pcnet_cs
604 12: 182918 XT-PIC PS/2 Mouse
606 14: 1232265 XT-PIC ide0
610 In 2.4.* a couple of lines where added to this file LOC & ERR (this time is the
611 output of a SMP machine):
613 > cat /proc/interrupts
616 0: 1243498 1214548 IO-APIC-edge timer
617 1: 8949 8958 IO-APIC-edge keyboard
618 2: 0 0 XT-PIC cascade
619 5: 11286 10161 IO-APIC-edge soundblaster
620 8: 1 0 IO-APIC-edge rtc
621 9: 27422 27407 IO-APIC-edge 3c503
622 12: 113645 113873 IO-APIC-edge PS/2 Mouse
624 14: 22491 24012 IO-APIC-edge ide0
625 15: 2183 2415 IO-APIC-edge ide1
626 17: 30564 30414 IO-APIC-level eth0
627 18: 177 164 IO-APIC-level bttv
632 NMI is incremented in this case because every timer interrupt generates a NMI
633 (Non Maskable Interrupt) which is used by the NMI Watchdog to detect lockups.
635 LOC is the local interrupt counter of the internal APIC of every CPU.
637 ERR is incremented in the case of errors in the IO-APIC bus (the bus that
638 connects the CPUs in a SMP system. This means that an error has been detected,
639 the IO-APIC automatically retry the transmission, so it should not be a big
640 problem, but you should read the SMP-FAQ.
642 In 2.6.2* /proc/interrupts was expanded again. This time the goal was for
643 /proc/interrupts to display every IRQ vector in use by the system, not
644 just those considered 'most important'. The new vectors are:
646 THR -- interrupt raised when a machine check threshold counter
647 (typically counting ECC corrected errors of memory or cache) exceeds
648 a configurable threshold. Only available on some systems.
650 TRM -- a thermal event interrupt occurs when a temperature threshold
651 has been exceeded for the CPU. This interrupt may also be generated
652 when the temperature drops back to normal.
654 SPU -- a spurious interrupt is some interrupt that was raised then lowered
655 by some IO device before it could be fully processed by the APIC. Hence
656 the APIC sees the interrupt but does not know what device it came from.
657 For this case the APIC will generate the interrupt with a IRQ vector
658 of 0xff. This might also be generated by chipset bugs.
660 RES, CAL, TLB -- rescheduling, call and TLB flush interrupts are
661 sent from one CPU to another per the needs of the OS. Typically,
662 their statistics are used by kernel developers and interested users to
663 determine the occurrence of interrupts of the given type.
665 The above IRQ vectors are displayed only when relevant. For example,
666 the threshold vector does not exist on x86_64 platforms. Others are
667 suppressed when the system is a uniprocessor. As of this writing, only
668 i386 and x86_64 platforms support the new IRQ vector displays.
670 Of some interest is the introduction of the /proc/irq directory to 2.4.
671 It could be used to set IRQ to CPU affinity, this means that you can "hook" an
672 IRQ to only one CPU, or to exclude a CPU of handling IRQs. The contents of the
673 irq subdir is one subdir for each IRQ, and two files; default_smp_affinity and
678 0 10 12 14 16 18 2 4 6 8 prof_cpu_mask
679 1 11 13 15 17 19 3 5 7 9 default_smp_affinity
683 smp_affinity is a bitmask, in which you can specify which CPUs can handle the
684 IRQ, you can set it by doing:
686 > echo 1 > /proc/irq/10/smp_affinity
688 This means that only the first CPU will handle the IRQ, but you can also echo
689 5 which means that only the first and fourth CPU can handle the IRQ.
691 The contents of each smp_affinity file is the same by default:
693 > cat /proc/irq/0/smp_affinity
696 There is an alternate interface, smp_affinity_list which allows specifying
697 a cpu range instead of a bitmask:
699 > cat /proc/irq/0/smp_affinity_list
702 The default_smp_affinity mask applies to all non-active IRQs, which are the
703 IRQs which have not yet been allocated/activated, and hence which lack a
704 /proc/irq/[0-9]* directory.
706 The node file on an SMP system shows the node to which the device using the IRQ
707 reports itself as being attached. This hardware locality information does not
708 include information about any possible driver locality preference.
710 prof_cpu_mask specifies which CPUs are to be profiled by the system wide
711 profiler. Default value is ffffffff (all cpus if there are only 32 of them).
713 The way IRQs are routed is handled by the IO-APIC, and it's Round Robin
714 between all the CPUs which are allowed to handle it. As usual the kernel has
715 more info than you and does a better job than you, so the defaults are the
716 best choice for almost everyone. [Note this applies only to those IO-APIC's
717 that support "Round Robin" interrupt distribution.]
719 There are three more important subdirectories in /proc: net, scsi, and sys.
720 The general rule is that the contents, or even the existence of these
721 directories, depend on your kernel configuration. If SCSI is not enabled, the
722 directory scsi may not exist. The same is true with the net, which is there
723 only when networking support is present in the running kernel.
725 The slabinfo file gives information about memory usage at the slab level.
726 Linux uses slab pools for memory management above page level in version 2.2.
727 Commonly used objects have their own slab pool (such as network buffers,
728 directory cache, and so on).
730 ..............................................................................
732 > cat /proc/buddyinfo
734 Node 0, zone DMA 0 4 5 4 4 3 ...
735 Node 0, zone Normal 1 0 0 1 101 8 ...
736 Node 0, zone HighMem 2 0 0 1 1 0 ...
738 External fragmentation is a problem under some workloads, and buddyinfo is a
739 useful tool for helping diagnose these problems. Buddyinfo will give you a
740 clue as to how big an area you can safely allocate, or why a previous
743 Each column represents the number of pages of a certain order which are
744 available. In this case, there are 0 chunks of 2^0*PAGE_SIZE available in
745 ZONE_DMA, 4 chunks of 2^1*PAGE_SIZE in ZONE_DMA, 101 chunks of 2^4*PAGE_SIZE
746 available in ZONE_NORMAL, etc...
748 More information relevant to external fragmentation can be found in
751 > cat /proc/pagetypeinfo
755 Free pages count per migrate type at order 0 1 2 3 4 5 6 7 8 9 10
756 Node 0, zone DMA, type Unmovable 0 0 0 1 1 1 1 1 1 1 0
757 Node 0, zone DMA, type Reclaimable 0 0 0 0 0 0 0 0 0 0 0
758 Node 0, zone DMA, type Movable 1 1 2 1 2 1 1 0 1 0 2
759 Node 0, zone DMA, type Reserve 0 0 0 0 0 0 0 0 0 1 0
760 Node 0, zone DMA, type Isolate 0 0 0 0 0 0 0 0 0 0 0
761 Node 0, zone DMA32, type Unmovable 103 54 77 1 1 1 11 8 7 1 9
762 Node 0, zone DMA32, type Reclaimable 0 0 2 1 0 0 0 0 1 0 0
763 Node 0, zone DMA32, type Movable 169 152 113 91 77 54 39 13 6 1 452
764 Node 0, zone DMA32, type Reserve 1 2 2 2 2 0 1 1 1 1 0
765 Node 0, zone DMA32, type Isolate 0 0 0 0 0 0 0 0 0 0 0
767 Number of blocks type Unmovable Reclaimable Movable Reserve Isolate
768 Node 0, zone DMA 2 0 5 1 0
769 Node 0, zone DMA32 41 6 967 2 0
771 Fragmentation avoidance in the kernel works by grouping pages of different
772 migrate types into the same contiguous regions of memory called page blocks.
773 A page block is typically the size of the default hugepage size e.g. 2MB on
774 X86-64. By keeping pages grouped based on their ability to move, the kernel
775 can reclaim pages within a page block to satisfy a high-order allocation.
777 The pagetypinfo begins with information on the size of a page block. It
778 then gives the same type of information as buddyinfo except broken down
779 by migrate-type and finishes with details on how many page blocks of each
782 If min_free_kbytes has been tuned correctly (recommendations made by hugeadm
783 from libhugetlbfs http://sourceforge.net/projects/libhugetlbfs/), one can
784 make an estimate of the likely number of huge pages that can be allocated
785 at a given point in time. All the "Movable" blocks should be allocatable
786 unless memory has been mlock()'d. Some of the Reclaimable blocks should
787 also be allocatable although a lot of filesystem metadata may have to be
788 reclaimed to achieve this.
790 ..............................................................................
794 Provides information about distribution and utilization of memory. This
795 varies by architecture and compile options. The following is from a
796 16GB PIII, which has highmem enabled. You may not have all of these fields.
800 MemTotal: 16344972 kB
802 MemAvailable: 14836172 kB
808 HighTotal: 15597528 kB
809 HighFree: 13629632 kB
819 SReclaimable: 159856 kB
820 SUnreclaim: 124508 kB
825 CommitLimit: 7669796 kB
826 Committed_AS: 100056 kB
827 VmallocTotal: 112216 kB
829 VmallocChunk: 111088 kB
830 AnonHugePages: 49152 kB
832 MemTotal: Total usable ram (i.e. physical ram minus a few reserved
833 bits and the kernel binary code)
834 MemFree: The sum of LowFree+HighFree
835 MemAvailable: An estimate of how much memory is available for starting new
836 applications, without swapping. Calculated from MemFree,
837 SReclaimable, the size of the file LRU lists, and the low
838 watermarks in each zone.
839 The estimate takes into account that the system needs some
840 page cache to function well, and that not all reclaimable
841 slab will be reclaimable, due to items being in use. The
842 impact of those factors will vary from system to system.
843 Buffers: Relatively temporary storage for raw disk blocks
844 shouldn't get tremendously large (20MB or so)
845 Cached: in-memory cache for files read from the disk (the
846 pagecache). Doesn't include SwapCached
847 SwapCached: Memory that once was swapped out, is swapped back in but
848 still also is in the swapfile (if memory is needed it
849 doesn't need to be swapped out AGAIN because it is already
850 in the swapfile. This saves I/O)
851 Active: Memory that has been used more recently and usually not
852 reclaimed unless absolutely necessary.
853 Inactive: Memory which has been less recently used. It is more
854 eligible to be reclaimed for other purposes
856 HighFree: Highmem is all memory above ~860MB of physical memory
857 Highmem areas are for use by userspace programs, or
858 for the pagecache. The kernel must use tricks to access
859 this memory, making it slower to access than lowmem.
861 LowFree: Lowmem is memory which can be used for everything that
862 highmem can be used for, but it is also available for the
863 kernel's use for its own data structures. Among many
864 other things, it is where everything from the Slab is
865 allocated. Bad things happen when you're out of lowmem.
866 SwapTotal: total amount of swap space available
867 SwapFree: Memory which has been evicted from RAM, and is temporarily
869 Dirty: Memory which is waiting to get written back to the disk
870 Writeback: Memory which is actively being written back to the disk
871 AnonPages: Non-file backed pages mapped into userspace page tables
872 AnonHugePages: Non-file backed huge pages mapped into userspace page tables
873 Mapped: files which have been mmaped, such as libraries
874 Slab: in-kernel data structures cache
875 SReclaimable: Part of Slab, that might be reclaimed, such as caches
876 SUnreclaim: Part of Slab, that cannot be reclaimed on memory pressure
877 PageTables: amount of memory dedicated to the lowest level of page
879 NFS_Unstable: NFS pages sent to the server, but not yet committed to stable
881 Bounce: Memory used for block device "bounce buffers"
882 WritebackTmp: Memory used by FUSE for temporary writeback buffers
883 CommitLimit: Based on the overcommit ratio ('vm.overcommit_ratio'),
884 this is the total amount of memory currently available to
885 be allocated on the system. This limit is only adhered to
886 if strict overcommit accounting is enabled (mode 2 in
887 'vm.overcommit_memory').
888 The CommitLimit is calculated with the following formula:
889 CommitLimit = ([total RAM pages] - [total huge TLB pages]) *
890 overcommit_ratio / 100 + [total swap pages]
891 For example, on a system with 1G of physical RAM and 7G
892 of swap with a `vm.overcommit_ratio` of 30 it would
893 yield a CommitLimit of 7.3G.
894 For more details, see the memory overcommit documentation
895 in vm/overcommit-accounting.
896 Committed_AS: The amount of memory presently allocated on the system.
897 The committed memory is a sum of all of the memory which
898 has been allocated by processes, even if it has not been
899 "used" by them as of yet. A process which malloc()'s 1G
900 of memory, but only touches 300M of it will show up as
901 using 1G. This 1G is memory which has been "committed" to
902 by the VM and can be used at any time by the allocating
903 application. With strict overcommit enabled on the system
904 (mode 2 in 'vm.overcommit_memory'),allocations which would
905 exceed the CommitLimit (detailed above) will not be permitted.
906 This is useful if one needs to guarantee that processes will
907 not fail due to lack of memory once that memory has been
908 successfully allocated.
909 VmallocTotal: total size of vmalloc memory area
910 VmallocUsed: amount of vmalloc area which is used
911 VmallocChunk: largest contiguous block of vmalloc area which is free
913 ..............................................................................
917 Provides information about vmalloced/vmaped areas. One line per area,
918 containing the virtual address range of the area, size in bytes,
919 caller information of the creator, and optional information depending
920 on the kind of area :
922 pages=nr number of pages
923 phys=addr if a physical address was specified
924 ioremap I/O mapping (ioremap() and friends)
925 vmalloc vmalloc() area
928 vpages buffer for pages pointers was vmalloced (huge area)
929 N<node>=nr (Only on NUMA kernels)
930 Number of pages allocated on memory node <node>
932 > cat /proc/vmallocinfo
933 0xffffc20000000000-0xffffc20000201000 2101248 alloc_large_system_hash+0x204 ...
934 /0x2c0 pages=512 vmalloc N0=128 N1=128 N2=128 N3=128
935 0xffffc20000201000-0xffffc20000302000 1052672 alloc_large_system_hash+0x204 ...
936 /0x2c0 pages=256 vmalloc N0=64 N1=64 N2=64 N3=64
937 0xffffc20000302000-0xffffc20000304000 8192 acpi_tb_verify_table+0x21/0x4f...
938 phys=7fee8000 ioremap
939 0xffffc20000304000-0xffffc20000307000 12288 acpi_tb_verify_table+0x21/0x4f...
940 phys=7fee7000 ioremap
941 0xffffc2000031d000-0xffffc2000031f000 8192 init_vdso_vars+0x112/0x210
942 0xffffc2000031f000-0xffffc2000032b000 49152 cramfs_uncompress_init+0x2e ...
943 /0x80 pages=11 vmalloc N0=3 N1=3 N2=2 N3=3
944 0xffffc2000033a000-0xffffc2000033d000 12288 sys_swapon+0x640/0xac0 ...
946 0xffffc20000347000-0xffffc2000034c000 20480 xt_alloc_table_info+0xfe ...
947 /0x130 [x_tables] pages=4 vmalloc N0=4
948 0xffffffffa0000000-0xffffffffa000f000 61440 sys_init_module+0xc27/0x1d00 ...
949 pages=14 vmalloc N2=14
950 0xffffffffa000f000-0xffffffffa0014000 20480 sys_init_module+0xc27/0x1d00 ...
952 0xffffffffa0014000-0xffffffffa0017000 12288 sys_init_module+0xc27/0x1d00 ...
954 0xffffffffa0017000-0xffffffffa0022000 45056 sys_init_module+0xc27/0x1d00 ...
955 pages=10 vmalloc N0=10
957 ..............................................................................
961 Provides counts of softirq handlers serviced since boot time, for each cpu.
966 TIMER: 27166 27120 27097 27034
971 SCHED: 27035 26983 26971 26746
973 RCU: 1678 1769 2178 2250
976 1.3 IDE devices in /proc/ide
977 ----------------------------
979 The subdirectory /proc/ide contains information about all IDE devices of which
980 the kernel is aware. There is one subdirectory for each IDE controller, the
981 file drivers and a link for each IDE device, pointing to the device directory
982 in the controller specific subtree.
984 The file drivers contains general information about the drivers used for the
987 > cat /proc/ide/drivers
988 ide-cdrom version 4.53
989 ide-disk version 1.08
991 More detailed information can be found in the controller specific
992 subdirectories. These are named ide0, ide1 and so on. Each of these
993 directories contains the files shown in table 1-6.
996 Table 1-6: IDE controller info in /proc/ide/ide?
997 ..............................................................................
999 channel IDE channel (0 or 1)
1000 config Configuration (only for PCI/IDE bridge)
1002 model Type/Chipset of IDE controller
1003 ..............................................................................
1005 Each device connected to a controller has a separate subdirectory in the
1006 controllers directory. The files listed in table 1-7 are contained in these
1010 Table 1-7: IDE device information
1011 ..............................................................................
1014 capacity Capacity of the medium (in 512Byte blocks)
1015 driver driver and version
1016 geometry physical and logical geometry
1017 identify device identify block
1019 model device identifier
1020 settings device setup
1021 smart_thresholds IDE disk management thresholds
1022 smart_values IDE disk management values
1023 ..............................................................................
1025 The most interesting file is settings. This file contains a nice overview of
1026 the drive parameters:
1028 # cat /proc/ide/ide0/hda/settings
1029 name value min max mode
1030 ---- ----- --- --- ----
1031 bios_cyl 526 0 65535 rw
1032 bios_head 255 0 255 rw
1033 bios_sect 63 0 63 rw
1034 breada_readahead 4 0 127 rw
1036 file_readahead 72 0 2097151 rw
1038 keepsettings 0 0 1 rw
1039 max_kb_per_request 122 1 127 rw
1043 pio_mode write-only 0 255 w
1049 1.4 Networking info in /proc/net
1050 --------------------------------
1052 The subdirectory /proc/net follows the usual pattern. Table 1-8 shows the
1053 additional values you get for IP version 6 if you configure the kernel to
1054 support this. Table 1-9 lists the files and their meaning.
1057 Table 1-8: IPv6 info in /proc/net
1058 ..............................................................................
1060 udp6 UDP sockets (IPv6)
1061 tcp6 TCP sockets (IPv6)
1062 raw6 Raw device statistics (IPv6)
1063 igmp6 IP multicast addresses, which this host joined (IPv6)
1064 if_inet6 List of IPv6 interface addresses
1065 ipv6_route Kernel routing table for IPv6
1066 rt6_stats Global IPv6 routing tables statistics
1067 sockstat6 Socket statistics (IPv6)
1068 snmp6 Snmp data (IPv6)
1069 ..............................................................................
1072 Table 1-9: Network info in /proc/net
1073 ..............................................................................
1075 arp Kernel ARP table
1076 dev network devices with statistics
1077 dev_mcast the Layer2 multicast groups a device is listening too
1078 (interface index, label, number of references, number of bound
1080 dev_stat network device status
1081 ip_fwchains Firewall chain linkage
1082 ip_fwnames Firewall chain names
1083 ip_masq Directory containing the masquerading tables
1084 ip_masquerade Major masquerading table
1085 netstat Network statistics
1086 raw raw device statistics
1087 route Kernel routing table
1088 rpc Directory containing rpc info
1089 rt_cache Routing cache
1091 sockstat Socket statistics
1094 unix UNIX domain sockets
1095 wireless Wireless interface data (Wavelan etc)
1096 igmp IP multicast addresses, which this host joined
1097 psched Global packet scheduler parameters.
1098 netlink List of PF_NETLINK sockets
1099 ip_mr_vifs List of multicast virtual interfaces
1100 ip_mr_cache List of multicast routing cache
1101 ..............................................................................
1103 You can use this information to see which network devices are available in
1104 your system and how much traffic was routed over those devices:
1107 Inter-|Receive |[...
1108 face |bytes packets errs drop fifo frame compressed multicast|[...
1109 lo: 908188 5596 0 0 0 0 0 0 [...
1110 ppp0:15475140 20721 410 0 0 410 0 0 [...
1111 eth0: 614530 7085 0 0 0 0 0 1 [...
1114 ...] bytes packets errs drop fifo colls carrier compressed
1115 ...] 908188 5596 0 0 0 0 0 0
1116 ...] 1375103 17405 0 0 0 0 0 0
1117 ...] 1703981 5535 0 0 0 3 0 0
1119 In addition, each Channel Bond interface has its own directory. For
1120 example, the bond0 device will have a directory called /proc/net/bond0/.
1121 It will contain information that is specific to that bond, such as the
1122 current slaves of the bond, the link status of the slaves, and how
1123 many times the slaves link has failed.
1128 If you have a SCSI host adapter in your system, you'll find a subdirectory
1129 named after the driver for this adapter in /proc/scsi. You'll also see a list
1130 of all recognized SCSI devices in /proc/scsi:
1132 >cat /proc/scsi/scsi
1134 Host: scsi0 Channel: 00 Id: 00 Lun: 00
1135 Vendor: IBM Model: DGHS09U Rev: 03E0
1136 Type: Direct-Access ANSI SCSI revision: 03
1137 Host: scsi0 Channel: 00 Id: 06 Lun: 00
1138 Vendor: PIONEER Model: CD-ROM DR-U06S Rev: 1.04
1139 Type: CD-ROM ANSI SCSI revision: 02
1142 The directory named after the driver has one file for each adapter found in
1143 the system. These files contain information about the controller, including
1144 the used IRQ and the IO address range. The amount of information shown is
1145 dependent on the adapter you use. The example shows the output for an Adaptec
1146 AHA-2940 SCSI adapter:
1148 > cat /proc/scsi/aic7xxx/0
1150 Adaptec AIC7xxx driver version: 5.1.19/3.2.4
1152 TCQ Enabled By Default : Disabled
1153 AIC7XXX_PROC_STATS : Disabled
1154 AIC7XXX_RESET_DELAY : 5
1155 Adapter Configuration:
1156 SCSI Adapter: Adaptec AHA-294X Ultra SCSI host adapter
1157 Ultra Wide Controller
1158 PCI MMAPed I/O Base: 0xeb001000
1159 Adapter SEEPROM Config: SEEPROM found and used.
1160 Adaptec SCSI BIOS: Enabled
1162 SCBs: Active 0, Max Active 2,
1163 Allocated 15, HW 16, Page 255
1165 BIOS Control Word: 0x18b6
1166 Adapter Control Word: 0x005b
1167 Extended Translation: Enabled
1168 Disconnect Enable Flags: 0xffff
1169 Ultra Enable Flags: 0x0001
1170 Tag Queue Enable Flags: 0x0000
1171 Ordered Queue Tag Flags: 0x0000
1172 Default Tag Queue Depth: 8
1173 Tagged Queue By Device array for aic7xxx host instance 0:
1174 {255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255}
1175 Actual queue depth per device for aic7xxx host instance 0:
1176 {1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1}
1179 Device using Wide/Sync transfers at 40.0 MByte/sec, offset 8
1180 Transinfo settings: current(12/8/1/0), goal(12/8/1/0), user(12/15/1/0)
1181 Total transfers 160151 (74577 reads and 85574 writes)
1183 Device using Narrow/Sync transfers at 5.0 MByte/sec, offset 15
1184 Transinfo settings: current(50/15/0/0), goal(50/15/0/0), user(50/15/0/0)
1185 Total transfers 0 (0 reads and 0 writes)
1188 1.6 Parallel port info in /proc/parport
1189 ---------------------------------------
1191 The directory /proc/parport contains information about the parallel ports of
1192 your system. It has one subdirectory for each port, named after the port
1195 These directories contain the four files shown in Table 1-10.
1198 Table 1-10: Files in /proc/parport
1199 ..............................................................................
1201 autoprobe Any IEEE-1284 device ID information that has been acquired.
1202 devices list of the device drivers using that port. A + will appear by the
1203 name of the device currently using the port (it might not appear
1205 hardware Parallel port's base address, IRQ line and DMA channel.
1206 irq IRQ that parport is using for that port. This is in a separate
1207 file to allow you to alter it by writing a new value in (IRQ
1209 ..............................................................................
1211 1.7 TTY info in /proc/tty
1212 -------------------------
1214 Information about the available and actually used tty's can be found in the
1215 directory /proc/tty.You'll find entries for drivers and line disciplines in
1216 this directory, as shown in Table 1-11.
1219 Table 1-11: Files in /proc/tty
1220 ..............................................................................
1222 drivers list of drivers and their usage
1223 ldiscs registered line disciplines
1224 driver/serial usage statistic and status of single tty lines
1225 ..............................................................................
1227 To see which tty's are currently in use, you can simply look into the file
1230 > cat /proc/tty/drivers
1231 pty_slave /dev/pts 136 0-255 pty:slave
1232 pty_master /dev/ptm 128 0-255 pty:master
1233 pty_slave /dev/ttyp 3 0-255 pty:slave
1234 pty_master /dev/pty 2 0-255 pty:master
1235 serial /dev/cua 5 64-67 serial:callout
1236 serial /dev/ttyS 4 64-67 serial
1237 /dev/tty0 /dev/tty0 4 0 system:vtmaster
1238 /dev/ptmx /dev/ptmx 5 2 system
1239 /dev/console /dev/console 5 1 system:console
1240 /dev/tty /dev/tty 5 0 system:/dev/tty
1241 unknown /dev/tty 4 1-63 console
1244 1.8 Miscellaneous kernel statistics in /proc/stat
1245 -------------------------------------------------
1247 Various pieces of information about kernel activity are available in the
1248 /proc/stat file. All of the numbers reported in this file are aggregates
1249 since the system first booted. For a quick look, simply cat the file:
1252 cpu 2255 34 2290 22625563 6290 127 456 0 0 0
1253 cpu0 1132 34 1441 11311718 3675 127 438 0 0 0
1254 cpu1 1123 0 849 11313845 2614 0 18 0 0 0
1255 intr 114930548 113199788 3 0 5 263 0 4 [... lots more numbers ...]
1261 softirq 183433 0 21755 12 39 1137 231 21459 2263
1263 The very first "cpu" line aggregates the numbers in all of the other "cpuN"
1264 lines. These numbers identify the amount of time the CPU has spent performing
1265 different kinds of work. Time units are in USER_HZ (typically hundredths of a
1266 second). The meanings of the columns are as follows, from left to right:
1268 - user: normal processes executing in user mode
1269 - nice: niced processes executing in user mode
1270 - system: processes executing in kernel mode
1271 - idle: twiddling thumbs
1272 - iowait: waiting for I/O to complete
1273 - irq: servicing interrupts
1274 - softirq: servicing softirqs
1275 - steal: involuntary wait
1276 - guest: running a normal guest
1277 - guest_nice: running a niced guest
1279 The "intr" line gives counts of interrupts serviced since boot time, for each
1280 of the possible system interrupts. The first column is the total of all
1281 interrupts serviced including unnumbered architecture specific interrupts;
1282 each subsequent column is the total for that particular numbered interrupt.
1283 Unnumbered interrupts are not shown, only summed into the total.
1285 The "ctxt" line gives the total number of context switches across all CPUs.
1287 The "btime" line gives the time at which the system booted, in seconds since
1290 The "processes" line gives the number of processes and threads created, which
1291 includes (but is not limited to) those created by calls to the fork() and
1292 clone() system calls.
1294 The "procs_running" line gives the total number of threads that are
1295 running or ready to run (i.e., the total number of runnable threads).
1297 The "procs_blocked" line gives the number of processes currently blocked,
1298 waiting for I/O to complete.
1300 The "softirq" line gives counts of softirqs serviced since boot time, for each
1301 of the possible system softirqs. The first column is the total of all
1302 softirqs serviced; each subsequent column is the total for that particular
1306 1.9 Ext4 file system parameters
1307 -------------------------------
1309 Information about mounted ext4 file systems can be found in
1310 /proc/fs/ext4. Each mounted filesystem will have a directory in
1311 /proc/fs/ext4 based on its device name (i.e., /proc/fs/ext4/hdc or
1312 /proc/fs/ext4/dm-0). The files in each per-device directory are shown
1313 in Table 1-12, below.
1315 Table 1-12: Files in /proc/fs/ext4/<devname>
1316 ..............................................................................
1318 mb_groups details of multiblock allocator buddy cache of free blocks
1319 ..............................................................................
1323 Shows registered system console lines.
1325 To see which character device lines are currently used for the system console
1326 /dev/console, you may simply look into the file /proc/consoles:
1328 > cat /proc/consoles
1334 device name of the device
1335 operations R = can do read operations
1336 W = can do write operations
1338 flags E = it is enabled
1339 C = it is preferred console
1340 B = it is primary boot console
1341 p = it is used for printk buffer
1342 b = it is not a TTY but a Braille device
1343 a = it is safe to use when cpu is offline
1344 major:minor major and minor number of the device separated by a colon
1346 ------------------------------------------------------------------------------
1348 ------------------------------------------------------------------------------
1349 The /proc file system serves information about the running system. It not only
1350 allows access to process data but also allows you to request the kernel status
1351 by reading files in the hierarchy.
1353 The directory structure of /proc reflects the types of information and makes
1354 it easy, if not obvious, where to look for specific data.
1355 ------------------------------------------------------------------------------
1357 ------------------------------------------------------------------------------
1358 CHAPTER 2: MODIFYING SYSTEM PARAMETERS
1359 ------------------------------------------------------------------------------
1361 ------------------------------------------------------------------------------
1363 ------------------------------------------------------------------------------
1364 * Modifying kernel parameters by writing into files found in /proc/sys
1365 * Exploring the files which modify certain parameters
1366 * Review of the /proc/sys file tree
1367 ------------------------------------------------------------------------------
1370 A very interesting part of /proc is the directory /proc/sys. This is not only
1371 a source of information, it also allows you to change parameters within the
1372 kernel. Be very careful when attempting this. You can optimize your system,
1373 but you can also cause it to crash. Never alter kernel parameters on a
1374 production system. Set up a development machine and test to make sure that
1375 everything works the way you want it to. You may have no alternative but to
1376 reboot the machine once an error has been made.
1378 To change a value, simply echo the new value into the file. An example is
1379 given below in the section on the file system data. You need to be root to do
1380 this. You can create your own boot script to perform this every time your
1383 The files in /proc/sys can be used to fine tune and monitor miscellaneous and
1384 general things in the operation of the Linux kernel. Since some of the files
1385 can inadvertently disrupt your system, it is advisable to read both
1386 documentation and source before actually making adjustments. In any case, be
1387 very careful when writing to any of these files. The entries in /proc may
1388 change slightly between the 2.1.* and the 2.2 kernel, so if there is any doubt
1389 review the kernel documentation in the directory /usr/src/linux/Documentation.
1390 This chapter is heavily based on the documentation included in the pre 2.2
1391 kernels, and became part of it in version 2.2.1 of the Linux kernel.
1393 Please see: Documentation/sysctl/ directory for descriptions of these
1396 ------------------------------------------------------------------------------
1398 ------------------------------------------------------------------------------
1399 Certain aspects of kernel behavior can be modified at runtime, without the
1400 need to recompile the kernel, or even to reboot the system. The files in the
1401 /proc/sys tree can not only be read, but also modified. You can use the echo
1402 command to write value into these files, thereby changing the default settings
1404 ------------------------------------------------------------------------------
1406 ------------------------------------------------------------------------------
1407 CHAPTER 3: PER-PROCESS PARAMETERS
1408 ------------------------------------------------------------------------------
1410 3.1 /proc/<pid>/oom_adj & /proc/<pid>/oom_score_adj- Adjust the oom-killer score
1411 --------------------------------------------------------------------------------
1413 These file can be used to adjust the badness heuristic used to select which
1414 process gets killed in out of memory conditions.
1416 The badness heuristic assigns a value to each candidate task ranging from 0
1417 (never kill) to 1000 (always kill) to determine which process is targeted. The
1418 units are roughly a proportion along that range of allowed memory the process
1419 may allocate from based on an estimation of its current memory and swap use.
1420 For example, if a task is using all allowed memory, its badness score will be
1421 1000. If it is using half of its allowed memory, its score will be 500.
1423 There is an additional factor included in the badness score: the current memory
1424 and swap usage is discounted by 3% for root processes.
1426 The amount of "allowed" memory depends on the context in which the oom killer
1427 was called. If it is due to the memory assigned to the allocating task's cpuset
1428 being exhausted, the allowed memory represents the set of mems assigned to that
1429 cpuset. If it is due to a mempolicy's node(s) being exhausted, the allowed
1430 memory represents the set of mempolicy nodes. If it is due to a memory
1431 limit (or swap limit) being reached, the allowed memory is that configured
1432 limit. Finally, if it is due to the entire system being out of memory, the
1433 allowed memory represents all allocatable resources.
1435 The value of /proc/<pid>/oom_score_adj is added to the badness score before it
1436 is used to determine which task to kill. Acceptable values range from -1000
1437 (OOM_SCORE_ADJ_MIN) to +1000 (OOM_SCORE_ADJ_MAX). This allows userspace to
1438 polarize the preference for oom killing either by always preferring a certain
1439 task or completely disabling it. The lowest possible value, -1000, is
1440 equivalent to disabling oom killing entirely for that task since it will always
1441 report a badness score of 0.
1443 Consequently, it is very simple for userspace to define the amount of memory to
1444 consider for each task. Setting a /proc/<pid>/oom_score_adj value of +500, for
1445 example, is roughly equivalent to allowing the remainder of tasks sharing the
1446 same system, cpuset, mempolicy, or memory controller resources to use at least
1447 50% more memory. A value of -500, on the other hand, would be roughly
1448 equivalent to discounting 50% of the task's allowed memory from being considered
1449 as scoring against the task.
1451 For backwards compatibility with previous kernels, /proc/<pid>/oom_adj may also
1452 be used to tune the badness score. Its acceptable values range from -16
1453 (OOM_ADJUST_MIN) to +15 (OOM_ADJUST_MAX) and a special value of -17
1454 (OOM_DISABLE) to disable oom killing entirely for that task. Its value is
1455 scaled linearly with /proc/<pid>/oom_score_adj.
1457 The value of /proc/<pid>/oom_score_adj may be reduced no lower than the last
1458 value set by a CAP_SYS_RESOURCE process. To reduce the value any lower
1459 requires CAP_SYS_RESOURCE.
1461 Caveat: when a parent task is selected, the oom killer will sacrifice any first
1462 generation children with separate address spaces instead, if possible. This
1463 avoids servers and important system daemons from being killed and loses the
1464 minimal amount of work.
1467 3.2 /proc/<pid>/oom_score - Display current oom-killer score
1468 -------------------------------------------------------------
1470 This file can be used to check the current score used by the oom-killer is for
1471 any given <pid>. Use it together with /proc/<pid>/oom_score_adj to tune which
1472 process should be killed in an out-of-memory situation.
1475 3.3 /proc/<pid>/io - Display the IO accounting fields
1476 -------------------------------------------------------
1478 This file contains IO statistics for each running process
1483 test:/tmp # dd if=/dev/zero of=/tmp/test.dat &
1486 test:/tmp # cat /proc/3828/io
1492 write_bytes: 323932160
1493 cancelled_write_bytes: 0
1502 I/O counter: chars read
1503 The number of bytes which this task has caused to be read from storage. This
1504 is simply the sum of bytes which this process passed to read() and pread().
1505 It includes things like tty IO and it is unaffected by whether or not actual
1506 physical disk IO was required (the read might have been satisfied from
1513 I/O counter: chars written
1514 The number of bytes which this task has caused, or shall cause to be written
1515 to disk. Similar caveats apply here as with rchar.
1521 I/O counter: read syscalls
1522 Attempt to count the number of read I/O operations, i.e. syscalls like read()
1529 I/O counter: write syscalls
1530 Attempt to count the number of write I/O operations, i.e. syscalls like
1531 write() and pwrite().
1537 I/O counter: bytes read
1538 Attempt to count the number of bytes which this process really did cause to
1539 be fetched from the storage layer. Done at the submit_bio() level, so it is
1540 accurate for block-backed filesystems. <please add status regarding NFS and
1541 CIFS at a later time>
1547 I/O counter: bytes written
1548 Attempt to count the number of bytes which this process caused to be sent to
1549 the storage layer. This is done at page-dirtying time.
1552 cancelled_write_bytes
1553 ---------------------
1555 The big inaccuracy here is truncate. If a process writes 1MB to a file and
1556 then deletes the file, it will in fact perform no writeout. But it will have
1557 been accounted as having caused 1MB of write.
1558 In other words: The number of bytes which this process caused to not happen,
1559 by truncating pagecache. A task can cause "negative" IO too. If this task
1560 truncates some dirty pagecache, some IO which another task has been accounted
1561 for (in its write_bytes) will not be happening. We _could_ just subtract that
1562 from the truncating task's write_bytes, but there is information loss in doing
1569 At its current implementation state, this is a bit racy on 32-bit machines: if
1570 process A reads process B's /proc/pid/io while process B is updating one of
1571 those 64-bit counters, process A could see an intermediate result.
1574 More information about this can be found within the taskstats documentation in
1575 Documentation/accounting.
1577 3.4 /proc/<pid>/coredump_filter - Core dump filtering settings
1578 ---------------------------------------------------------------
1579 When a process is dumped, all anonymous memory is written to a core file as
1580 long as the size of the core file isn't limited. But sometimes we don't want
1581 to dump some memory segments, for example, huge shared memory or DAX.
1582 Conversely, sometimes we want to save file-backed memory segments into a core
1583 file, not only the individual files.
1585 /proc/<pid>/coredump_filter allows you to customize which memory segments
1586 will be dumped when the <pid> process is dumped. coredump_filter is a bitmask
1587 of memory types. If a bit of the bitmask is set, memory segments of the
1588 corresponding memory type are dumped, otherwise they are not dumped.
1590 The following 9 memory types are supported:
1591 - (bit 0) anonymous private memory
1592 - (bit 1) anonymous shared memory
1593 - (bit 2) file-backed private memory
1594 - (bit 3) file-backed shared memory
1595 - (bit 4) ELF header pages in file-backed private memory areas (it is
1596 effective only if the bit 2 is cleared)
1597 - (bit 5) hugetlb private memory
1598 - (bit 6) hugetlb shared memory
1599 - (bit 7) DAX private memory
1600 - (bit 8) DAX shared memory
1602 Note that MMIO pages such as frame buffer are never dumped and vDSO pages
1603 are always dumped regardless of the bitmask status.
1605 Note that bits 0-4 don't affect hugetlb or DAX memory. hugetlb memory is
1606 only affected by bit 5-6, and DAX is only affected by bits 7-8.
1608 The default value of coredump_filter is 0x33; this means all anonymous memory
1609 segments, ELF header pages and hugetlb private memory are dumped.
1611 If you don't want to dump all shared memory segments attached to pid 1234,
1612 write 0x31 to the process's proc file.
1614 $ echo 0x31 > /proc/1234/coredump_filter
1616 When a new process is created, the process inherits the bitmask status from its
1617 parent. It is useful to set up coredump_filter before the program runs.
1620 $ echo 0x7 > /proc/self/coredump_filter
1623 3.5 /proc/<pid>/mountinfo - Information about mounts
1624 --------------------------------------------------------
1626 This file contains lines of the form:
1628 36 35 98:0 /mnt1 /mnt2 rw,noatime master:1 - ext3 /dev/root rw,errors=continue
1629 (1)(2)(3) (4) (5) (6) (7) (8) (9) (10) (11)
1631 (1) mount ID: unique identifier of the mount (may be reused after umount)
1632 (2) parent ID: ID of parent (or of self for the top of the mount tree)
1633 (3) major:minor: value of st_dev for files on filesystem
1634 (4) root: root of the mount within the filesystem
1635 (5) mount point: mount point relative to the process's root
1636 (6) mount options: per mount options
1637 (7) optional fields: zero or more fields of the form "tag[:value]"
1638 (8) separator: marks the end of the optional fields
1639 (9) filesystem type: name of filesystem of the form "type[.subtype]"
1640 (10) mount source: filesystem specific information or "none"
1641 (11) super options: per super block options
1643 Parsers should ignore all unrecognised optional fields. Currently the
1644 possible optional fields are:
1646 shared:X mount is shared in peer group X
1647 master:X mount is slave to peer group X
1648 propagate_from:X mount is slave and receives propagation from peer group X (*)
1649 unbindable mount is unbindable
1651 (*) X is the closest dominant peer group under the process's root. If
1652 X is the immediate master of the mount, or if there's no dominant peer
1653 group under the same root, then only the "master:X" field is present
1654 and not the "propagate_from:X" field.
1656 For more information on mount propagation see:
1658 Documentation/filesystems/sharedsubtree.txt
1661 3.6 /proc/<pid>/comm & /proc/<pid>/task/<tid>/comm
1662 --------------------------------------------------------
1663 These files provide a method to access a tasks comm value. It also allows for
1664 a task to set its own or one of its thread siblings comm value. The comm value
1665 is limited in size compared to the cmdline value, so writing anything longer
1666 then the kernel's TASK_COMM_LEN (currently 16 chars) will result in a truncated
1670 3.7 /proc/<pid>/task/<tid>/children - Information about task children
1671 -------------------------------------------------------------------------
1672 This file provides a fast way to retrieve first level children pids
1673 of a task pointed by <pid>/<tid> pair. The format is a space separated
1676 Note the "first level" here -- if a child has own children they will
1677 not be listed here, one needs to read /proc/<children-pid>/task/<tid>/children
1678 to obtain the descendants.
1680 Since this interface is intended to be fast and cheap it doesn't
1681 guarantee to provide precise results and some children might be
1682 skipped, especially if they've exited right after we printed their
1683 pids, so one need to either stop or freeze processes being inspected
1684 if precise results are needed.
1687 3.8 /proc/<pid>/fdinfo/<fd> - Information about opened file
1688 ---------------------------------------------------------------
1689 This file provides information associated with an opened file. The regular
1690 files have at least three fields -- 'pos', 'flags' and mnt_id. The 'pos'
1691 represents the current offset of the opened file in decimal form [see lseek(2)
1692 for details], 'flags' denotes the octal O_xxx mask the file has been
1693 created with [see open(2) for details] and 'mnt_id' represents mount ID of
1694 the file system containing the opened file [see 3.5 /proc/<pid>/mountinfo
1703 All locks associated with a file descriptor are shown in its fdinfo too.
1705 lock: 1: FLOCK ADVISORY WRITE 359 00:13:11691 0 EOF
1707 The files such as eventfd, fsnotify, signalfd, epoll among the regular pos/flags
1708 pair provide additional information particular to the objects they represent.
1717 where 'eventfd-count' is hex value of a counter.
1724 sigmask: 0000000000000200
1726 where 'sigmask' is hex value of the signal mask associated
1734 tfd: 5 events: 1d data: ffffffffffffffff
1736 where 'tfd' is a target file descriptor number in decimal form,
1737 'events' is events mask being watched and the 'data' is data
1738 associated with a target [see epoll(7) for more details].
1742 For inotify files the format is the following
1746 inotify wd:3 ino:9e7e sdev:800013 mask:800afce ignored_mask:0 fhandle-bytes:8 fhandle-type:1 f_handle:7e9e0000640d1b6d
1748 where 'wd' is a watch descriptor in decimal form, ie a target file
1749 descriptor number, 'ino' and 'sdev' are inode and device where the
1750 target file resides and the 'mask' is the mask of events, all in hex
1751 form [see inotify(7) for more details].
1753 If the kernel was built with exportfs support, the path to the target
1754 file is encoded as a file handle. The file handle is provided by three
1755 fields 'fhandle-bytes', 'fhandle-type' and 'f_handle', all in hex
1758 If the kernel is built without exportfs support the file handle won't be
1761 If there is no inotify mark attached yet the 'inotify' line will be omitted.
1763 For fanotify files the format is
1768 fanotify flags:10 event-flags:0
1769 fanotify mnt_id:12 mflags:40 mask:38 ignored_mask:40000003
1770 fanotify ino:4f969 sdev:800013 mflags:0 mask:3b ignored_mask:40000000 fhandle-bytes:8 fhandle-type:1 f_handle:69f90400c275b5b4
1772 where fanotify 'flags' and 'event-flags' are values used in fanotify_init
1773 call, 'mnt_id' is the mount point identifier, 'mflags' is the value of
1774 flags associated with mark which are tracked separately from events
1775 mask. 'ino', 'sdev' are target inode and device, 'mask' is the events
1776 mask and 'ignored_mask' is the mask of events which are to be ignored.
1777 All in hex format. Incorporation of 'mflags', 'mask' and 'ignored_mask'
1778 does provide information about flags and mask used in fanotify_mark
1779 call [see fsnotify manpage for details].
1781 While the first three lines are mandatory and always printed, the rest is
1782 optional and may be omitted if no marks created yet.
1793 it_value: (0, 49406829)
1796 where 'clockid' is the clock type and 'ticks' is the number of the timer expirations
1797 that have occurred [see timerfd_create(2) for details]. 'settime flags' are
1798 flags in octal form been used to setup the timer [see timerfd_settime(2) for
1799 details]. 'it_value' is remaining time until the timer exiration.
1800 'it_interval' is the interval for the timer. Note the timer might be set up
1801 with TIMER_ABSTIME option which will be shown in 'settime flags', but 'it_value'
1802 still exhibits timer's remaining time.
1804 3.9 /proc/<pid>/map_files - Information about memory mapped files
1805 ---------------------------------------------------------------------
1806 This directory contains symbolic links which represent memory mapped files
1807 the process is maintaining. Example output:
1809 | lr-------- 1 root root 64 Jan 27 11:24 333c600000-333c620000 -> /usr/lib64/ld-2.18.so
1810 | lr-------- 1 root root 64 Jan 27 11:24 333c81f000-333c820000 -> /usr/lib64/ld-2.18.so
1811 | lr-------- 1 root root 64 Jan 27 11:24 333c820000-333c821000 -> /usr/lib64/ld-2.18.so
1813 | lr-------- 1 root root 64 Jan 27 11:24 35d0421000-35d0422000 -> /usr/lib64/libselinux.so.1
1814 | lr-------- 1 root root 64 Jan 27 11:24 400000-41a000 -> /usr/bin/ls
1816 The name of a link represents the virtual memory bounds of a mapping, i.e.
1817 vm_area_struct::vm_start-vm_area_struct::vm_end.
1819 The main purpose of the map_files is to retrieve a set of memory mapped
1820 files in a fast way instead of parsing /proc/<pid>/maps or
1821 /proc/<pid>/smaps, both of which contain many more records. At the same
1822 time one can open(2) mappings from the listings of two processes and
1823 comparing their inode numbers to figure out which anonymous memory areas
1824 are actually shared.
1826 ------------------------------------------------------------------------------
1828 ------------------------------------------------------------------------------
1831 ---------------------
1833 The following mount options are supported:
1835 hidepid= Set /proc/<pid>/ access mode.
1836 gid= Set the group authorized to learn processes information.
1838 hidepid=0 means classic mode - everybody may access all /proc/<pid>/ directories
1841 hidepid=1 means users may not access any /proc/<pid>/ directories but their
1842 own. Sensitive files like cmdline, sched*, status are now protected against
1843 other users. This makes it impossible to learn whether any user runs
1844 specific program (given the program doesn't reveal itself by its behaviour).
1845 As an additional bonus, as /proc/<pid>/cmdline is unaccessible for other users,
1846 poorly written programs passing sensitive information via program arguments are
1847 now protected against local eavesdroppers.
1849 hidepid=2 means hidepid=1 plus all /proc/<pid>/ will be fully invisible to other
1850 users. It doesn't mean that it hides a fact whether a process with a specific
1851 pid value exists (it can be learned by other means, e.g. by "kill -0 $PID"),
1852 but it hides process' uid and gid, which may be learned by stat()'ing
1853 /proc/<pid>/ otherwise. It greatly complicates an intruder's task of gathering
1854 information about running processes, whether some daemon runs with elevated
1855 privileges, whether other user runs some sensitive program, whether other users
1856 run any program at all, etc.
1858 gid= defines a group authorized to learn processes information otherwise
1859 prohibited by hidepid=. If you use some daemon like identd which needs to learn
1860 information about processes information, just add identd to this group.