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
49 ------------------------------------------------------------------------------
51 ------------------------------------------------------------------------------
53 0.1 Introduction/Credits
54 ------------------------
56 This documentation is part of a soon (or so we hope) to be released book on
57 the SuSE Linux distribution. As there is no complete documentation for the
58 /proc file system and we've used many freely available sources to write these
59 chapters, it seems only fair to give the work back to the Linux community.
60 This work is based on the 2.2.* kernel version and the upcoming 2.4.*. I'm
61 afraid it's still far from complete, but we hope it will be useful. As far as
62 we know, it is the first 'all-in-one' document about the /proc file system. It
63 is focused on the Intel x86 hardware, so if you are looking for PPC, ARM,
64 SPARC, AXP, etc., features, you probably won't find what you are looking for.
65 It also only covers IPv4 networking, not IPv6 nor other protocols - sorry. But
66 additions and patches are welcome and will be added to this document if you
69 We'd like to thank Alan Cox, Rik van Riel, and Alexey Kuznetsov and a lot of
70 other people for help compiling this documentation. We'd also like to extend a
71 special thank you to Andi Kleen for documentation, which we relied on heavily
72 to create this document, as well as the additional information he provided.
73 Thanks to everybody else who contributed source or docs to the Linux kernel
74 and helped create a great piece of software... :)
76 If you have any comments, corrections or additions, please don't hesitate to
77 contact Bodo Bauer at bb@ricochet.net. We'll be happy to add them to this
80 The latest version of this document is available online at
81 http://tldp.org/LDP/Linux-Filesystem-Hierarchy/html/proc.html
83 If the above direction does not works for you, you could try the kernel
84 mailing list at linux-kernel@vger.kernel.org and/or try to reach me at
85 comandante@zaralinux.com.
90 We don't guarantee the correctness of this document, and if you come to us
91 complaining about how you screwed up your system because of incorrect
92 documentation, we won't feel responsible...
94 ------------------------------------------------------------------------------
95 CHAPTER 1: COLLECTING SYSTEM INFORMATION
96 ------------------------------------------------------------------------------
98 ------------------------------------------------------------------------------
100 ------------------------------------------------------------------------------
101 * Investigating the properties of the pseudo file system /proc and its
102 ability to provide information on the running Linux system
103 * Examining /proc's structure
104 * Uncovering various information about the kernel and the processes running
106 ------------------------------------------------------------------------------
109 The proc file system acts as an interface to internal data structures in the
110 kernel. It can be used to obtain information about the system and to change
111 certain kernel parameters at runtime (sysctl).
113 First, we'll take a look at the read-only parts of /proc. In Chapter 2, we
114 show you how you can use /proc/sys to change settings.
116 1.1 Process-Specific Subdirectories
117 -----------------------------------
119 The directory /proc contains (among other things) one subdirectory for each
120 process running on the system, which is named after the process ID (PID).
122 The link self points to the process reading the file system. Each process
123 subdirectory has the entries listed in Table 1-1.
126 Table 1-1: Process specific entries in /proc
127 ..............................................................................
129 clear_refs Clears page referenced bits shown in smaps output
130 cmdline Command line arguments
131 cpu Current and last cpu in which it was executed (2.4)(smp)
132 cwd Link to the current working directory
133 environ Values of environment variables
134 exe Link to the executable of this process
135 fd Directory, which contains all file descriptors
136 maps Memory maps to executables and library files (2.4)
137 mem Memory held by this process
138 root Link to the root directory of this process
140 statm Process memory status information
141 status Process status in human readable form
142 wchan If CONFIG_KALLSYMS is set, a pre-decoded wchan
144 stack Report full stack trace, enable via CONFIG_STACKTRACE
145 smaps a extension based on maps, showing the memory consumption of
146 each mapping and flags associated with it
147 ..............................................................................
149 For example, to get the status information of a process, all you have to do is
150 read the file /proc/PID/status:
152 >cat /proc/self/status
176 SigPnd: 0000000000000000
177 ShdPnd: 0000000000000000
178 SigBlk: 0000000000000000
179 SigIgn: 0000000000000000
180 SigCgt: 0000000000000000
181 CapInh: 00000000fffffeff
182 CapPrm: 0000000000000000
183 CapEff: 0000000000000000
184 CapBnd: ffffffffffffffff
186 voluntary_ctxt_switches: 0
187 nonvoluntary_ctxt_switches: 1
189 This shows you nearly the same information you would get if you viewed it with
190 the ps command. In fact, ps uses the proc file system to obtain its
191 information. But you get a more detailed view of the process by reading the
192 file /proc/PID/status. It fields are described in table 1-2.
194 The statm file contains more detailed information about the process
195 memory usage. Its seven fields are explained in Table 1-3. The stat file
196 contains details information about the process itself. Its fields are
197 explained in Table 1-4.
199 (for SMP CONFIG users)
200 For making accounting scalable, RSS related information are handled in
201 asynchronous manner and the vaule may not be very precise. To see a precise
202 snapshot of a moment, you can see /proc/<pid>/smaps file and scan page table.
203 It's slow but very precise.
205 Table 1-2: Contents of the status files (as of 2.6.30-rc7)
206 ..............................................................................
208 Name filename of the executable
209 State state (R is running, S is sleeping, D is sleeping
210 in an uninterruptible wait, Z is zombie,
211 T is traced or stopped)
214 PPid process id of the parent process
215 TracerPid PID of process tracing this process (0 if not)
216 Uid Real, effective, saved set, and file system UIDs
217 Gid Real, effective, saved set, and file system GIDs
218 FDSize number of file descriptor slots currently allocated
219 Groups supplementary group list
220 VmPeak peak virtual memory size
221 VmSize total program size
222 VmLck locked memory size
223 VmHWM peak resident set size ("high water mark")
224 VmRSS size of memory portions
225 VmData size of data, stack, and text segments
226 VmStk size of data, stack, and text segments
227 VmExe size of text segment
228 VmLib size of shared library code
229 VmPTE size of page table entries
230 VmSwap size of swap usage (the number of referred swapents)
231 Threads number of threads
232 SigQ number of signals queued/max. number for queue
233 SigPnd bitmap of pending signals for the thread
234 ShdPnd bitmap of shared pending signals for the process
235 SigBlk bitmap of blocked signals
236 SigIgn bitmap of ignored signals
237 SigCgt bitmap of catched signals
238 CapInh bitmap of inheritable capabilities
239 CapPrm bitmap of permitted capabilities
240 CapEff bitmap of effective capabilities
241 CapBnd bitmap of capabilities bounding set
242 Seccomp seccomp mode, like prctl(PR_GET_SECCOMP, ...)
243 Cpus_allowed mask of CPUs on which this process may run
244 Cpus_allowed_list Same as previous, but in "list format"
245 Mems_allowed mask of memory nodes allowed to this process
246 Mems_allowed_list Same as previous, but in "list format"
247 voluntary_ctxt_switches number of voluntary context switches
248 nonvoluntary_ctxt_switches number of non voluntary context switches
249 ..............................................................................
251 Table 1-3: Contents of the statm files (as of 2.6.8-rc3)
252 ..............................................................................
254 size total program size (pages) (same as VmSize in status)
255 resident size of memory portions (pages) (same as VmRSS in status)
256 shared number of pages that are shared (i.e. backed by a file)
257 trs number of pages that are 'code' (not including libs; broken,
258 includes data segment)
259 lrs number of pages of library (always 0 on 2.6)
260 drs number of pages of data/stack (including libs; broken,
261 includes library text)
262 dt number of dirty pages (always 0 on 2.6)
263 ..............................................................................
266 Table 1-4: Contents of the stat files (as of 2.6.30-rc7)
267 ..............................................................................
270 tcomm filename of the executable
271 state state (R is running, S is sleeping, D is sleeping in an
272 uninterruptible wait, Z is zombie, T is traced or stopped)
273 ppid process id of the parent process
274 pgrp pgrp of the process
276 tty_nr tty the process uses
277 tty_pgrp pgrp of the tty
279 min_flt number of minor faults
280 cmin_flt number of minor faults with child's
281 maj_flt number of major faults
282 cmaj_flt number of major faults with child's
283 utime user mode jiffies
284 stime kernel mode jiffies
285 cutime user mode jiffies with child's
286 cstime kernel mode jiffies with child's
287 priority priority level
289 num_threads number of threads
290 it_real_value (obsolete, always 0)
291 start_time time the process started after system boot
292 vsize virtual memory size
293 rss resident set memory size
294 rsslim current limit in bytes on the rss
295 start_code address above which program text can run
296 end_code address below which program text can run
297 start_stack address of the start of the main process stack
298 esp current value of ESP
299 eip current value of EIP
300 pending bitmap of pending signals
301 blocked bitmap of blocked signals
302 sigign bitmap of ignored signals
303 sigcatch bitmap of catched signals
304 wchan address where process went to sleep
307 exit_signal signal to send to parent thread on exit
308 task_cpu which CPU the task is scheduled on
309 rt_priority realtime priority
310 policy scheduling policy (man sched_setscheduler)
311 blkio_ticks time spent waiting for block IO
312 gtime guest time of the task in jiffies
313 cgtime guest time of the task children in jiffies
314 start_data address above which program data+bss is placed
315 end_data address below which program data+bss is placed
316 start_brk address above which program heap can be expanded with brk()
317 arg_start address above which program command line is placed
318 arg_end address below which program command line is placed
319 env_start address above which program environment is placed
320 env_end address below which program environment is placed
321 exit_code the thread's exit_code in the form reported by the waitpid system call
322 ..............................................................................
324 The /proc/PID/maps file containing the currently mapped memory regions and
325 their access permissions.
329 address perms offset dev inode pathname
331 08048000-08049000 r-xp 00000000 03:00 8312 /opt/test
332 08049000-0804a000 rw-p 00001000 03:00 8312 /opt/test
333 0804a000-0806b000 rw-p 00000000 00:00 0 [heap]
334 a7cb1000-a7cb2000 ---p 00000000 00:00 0
335 a7cb2000-a7eb2000 rw-p 00000000 00:00 0
336 a7eb2000-a7eb3000 ---p 00000000 00:00 0
337 a7eb3000-a7ed5000 rw-p 00000000 00:00 0 [stack:1001]
338 a7ed5000-a8008000 r-xp 00000000 03:00 4222 /lib/libc.so.6
339 a8008000-a800a000 r--p 00133000 03:00 4222 /lib/libc.so.6
340 a800a000-a800b000 rw-p 00135000 03:00 4222 /lib/libc.so.6
341 a800b000-a800e000 rw-p 00000000 00:00 0
342 a800e000-a8022000 r-xp 00000000 03:00 14462 /lib/libpthread.so.0
343 a8022000-a8023000 r--p 00013000 03:00 14462 /lib/libpthread.so.0
344 a8023000-a8024000 rw-p 00014000 03:00 14462 /lib/libpthread.so.0
345 a8024000-a8027000 rw-p 00000000 00:00 0
346 a8027000-a8043000 r-xp 00000000 03:00 8317 /lib/ld-linux.so.2
347 a8043000-a8044000 r--p 0001b000 03:00 8317 /lib/ld-linux.so.2
348 a8044000-a8045000 rw-p 0001c000 03:00 8317 /lib/ld-linux.so.2
349 aff35000-aff4a000 rw-p 00000000 00:00 0 [stack]
350 ffffe000-fffff000 r-xp 00000000 00:00 0 [vdso]
352 where "address" is the address space in the process that it occupies, "perms"
353 is a set of permissions:
359 p = private (copy on write)
361 "offset" is the offset into the mapping, "dev" is the device (major:minor), and
362 "inode" is the inode on that device. 0 indicates that no inode is associated
363 with the memory region, as the case would be with BSS (uninitialized data).
364 The "pathname" shows the name associated file for this mapping. If the mapping
365 is not associated with a file:
367 [heap] = the heap of the program
368 [stack] = the stack of the main process
369 [stack:1001] = the stack of the thread with tid 1001
370 [vdso] = the "virtual dynamic shared object",
371 the kernel system call handler
372 [anon:<name>] = an anonymous mapping that has been
375 or if empty, the mapping is anonymous.
377 The /proc/PID/task/TID/maps is a view of the virtual memory from the viewpoint
378 of the individual tasks of a process. In this file you will see a mapping marked
379 as [stack] if that task sees it as a stack. This is a key difference from the
380 content of /proc/PID/maps, where you will see all mappings that are being used
381 as stack by all of those tasks. Hence, for the example above, the task-level
382 map, i.e. /proc/PID/task/TID/maps for thread 1001 will look like this:
384 08048000-08049000 r-xp 00000000 03:00 8312 /opt/test
385 08049000-0804a000 rw-p 00001000 03:00 8312 /opt/test
386 0804a000-0806b000 rw-p 00000000 00:00 0 [heap]
387 a7cb1000-a7cb2000 ---p 00000000 00:00 0
388 a7cb2000-a7eb2000 rw-p 00000000 00:00 0
389 a7eb2000-a7eb3000 ---p 00000000 00:00 0
390 a7eb3000-a7ed5000 rw-p 00000000 00:00 0 [stack]
391 a7ed5000-a8008000 r-xp 00000000 03:00 4222 /lib/libc.so.6
392 a8008000-a800a000 r--p 00133000 03:00 4222 /lib/libc.so.6
393 a800a000-a800b000 rw-p 00135000 03:00 4222 /lib/libc.so.6
394 a800b000-a800e000 rw-p 00000000 00:00 0
395 a800e000-a8022000 r-xp 00000000 03:00 14462 /lib/libpthread.so.0
396 a8022000-a8023000 r--p 00013000 03:00 14462 /lib/libpthread.so.0
397 a8023000-a8024000 rw-p 00014000 03:00 14462 /lib/libpthread.so.0
398 a8024000-a8027000 rw-p 00000000 00:00 0
399 a8027000-a8043000 r-xp 00000000 03:00 8317 /lib/ld-linux.so.2
400 a8043000-a8044000 r--p 0001b000 03:00 8317 /lib/ld-linux.so.2
401 a8044000-a8045000 rw-p 0001c000 03:00 8317 /lib/ld-linux.so.2
402 aff35000-aff4a000 rw-p 00000000 00:00 0
403 ffffe000-fffff000 r-xp 00000000 00:00 0 [vdso]
405 The /proc/PID/smaps is an extension based on maps, showing the memory
406 consumption for each of the process's mappings. For each of mappings there
407 is a series of lines such as the following:
409 08048000-080bc000 r-xp 00000000 03:02 13130 /bin/bash
423 VmFlags: rd ex mr mw me de
424 Name: name from userspace
426 the first of these lines shows the same information as is displayed for the
427 mapping in /proc/PID/maps. The remaining lines show the size of the mapping
428 (size), the amount of the mapping that is currently resident in RAM (RSS), the
429 process' proportional share of this mapping (PSS), the number of clean and
430 dirty private pages in the mapping. Note that even a page which is part of a
431 MAP_SHARED mapping, but has only a single pte mapped, i.e. is currently used
432 by only one process, is accounted as private and not as shared. "Referenced"
433 indicates the amount of memory currently marked as referenced or accessed.
434 "Anonymous" shows the amount of memory that does not belong to any file. Even
435 a mapping associated with a file may contain anonymous pages: when MAP_PRIVATE
436 and a page is modified, the file page is replaced by a private anonymous copy.
437 "Swap" shows how much would-be-anonymous memory is also used, but out on
440 "VmFlags" field deserves a separate description. This member represents the kernel
441 flags associated with the particular virtual memory area in two letter encoded
442 manner. The codes are the following:
451 gd - stack segment growns down
453 dw - disabled write to the mapped file
454 lo - pages are locked in memory
455 io - memory mapped I/O area
456 sr - sequential read advise provided
457 rr - random read advise provided
458 dc - do not copy area on fork
459 de - do not expand area on remapping
460 ac - area is accountable
461 nr - swap space is not reserved for the area
462 ht - area uses huge tlb pages
463 nl - non-linear mapping
464 ar - architecture specific flag
465 dd - do not include area into core dump
467 hg - huge page advise flag
468 nh - no-huge page advise flag
469 mg - mergable advise flag
471 Note that there is no guarantee that every flag and associated mnemonic will
472 be present in all further kernel releases. Things get changed, the flags may
473 be vanished or the reverse -- new added.
475 The "Name" field will only be present on a mapping that has been named by
476 userspace, and will show the name passed in by userspace.
478 This file is only present if the CONFIG_MMU kernel configuration option is
481 The /proc/PID/clear_refs is used to reset the PG_Referenced and ACCESSED/YOUNG
482 bits on both physical and virtual pages associated with a process.
483 To clear the bits for all the pages associated with the process
484 > echo 1 > /proc/PID/clear_refs
486 To clear the bits for the anonymous pages associated with the process
487 > echo 2 > /proc/PID/clear_refs
489 To clear the bits for the file mapped pages associated with the process
490 > echo 3 > /proc/PID/clear_refs
491 Any other value written to /proc/PID/clear_refs will have no effect.
493 To reset the peak resident set size ("high water mark") to the process's
495 > echo 5 > /proc/PID/clear_refs
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.
504 Similar to the process entries, the kernel data files give information about
505 the running kernel. The files used to obtain this information are contained in
506 /proc and are listed in Table 1-5. Not all of these will be present in your
507 system. It depends on the kernel configuration and the loaded modules, which
508 files are there, and which are missing.
510 Table 1-5: Kernel info in /proc
511 ..............................................................................
513 apm Advanced power management info
514 buddyinfo Kernel memory allocator information (see text) (2.5)
515 bus Directory containing bus specific information
516 cmdline Kernel command line
517 cpuinfo Info about the CPU
518 devices Available devices (block and character)
519 dma Used DMS channels
520 filesystems Supported filesystems
521 driver Various drivers grouped here, currently rtc (2.4)
522 execdomains Execdomains, related to security (2.4)
523 fb Frame Buffer devices (2.4)
524 fs File system parameters, currently nfs/exports (2.4)
525 ide Directory containing info about the IDE subsystem
526 interrupts Interrupt usage
527 iomem Memory map (2.4)
528 ioports I/O port usage
529 irq Masks for irq to cpu affinity (2.4)(smp?)
530 isapnp ISA PnP (Plug&Play) Info (2.4)
531 kcore Kernel core image (can be ELF or A.OUT(deprecated in 2.4))
533 ksyms Kernel symbol table
534 loadavg Load average of last 1, 5 & 15 minutes
538 modules List of loaded modules
539 mounts Mounted filesystems
540 net Networking info (see text)
541 pagetypeinfo Additional page allocator information (see text) (2.5)
542 partitions Table of partitions known to the system
543 pci Deprecated info of PCI bus (new way -> /proc/bus/pci/,
544 decoupled by lspci (2.4)
546 scsi SCSI info (see text)
547 slabinfo Slab pool info
548 softirqs softirq usage
549 stat Overall statistics
550 swaps Swap space utilization
552 sysvipc Info of SysVIPC Resources (msg, sem, shm) (2.4)
553 tty Info of tty drivers
555 version Kernel version
556 video bttv info of video resources (2.4)
557 vmallocinfo Show vmalloced areas
558 ..............................................................................
560 You can, for example, check which interrupts are currently in use and what
561 they are used for by looking in the file /proc/interrupts:
563 > cat /proc/interrupts
565 0: 8728810 XT-PIC timer
566 1: 895 XT-PIC keyboard
568 3: 531695 XT-PIC aha152x
569 4: 2014133 XT-PIC serial
570 5: 44401 XT-PIC pcnet_cs
573 12: 182918 XT-PIC PS/2 Mouse
575 14: 1232265 XT-PIC ide0
579 In 2.4.* a couple of lines where added to this file LOC & ERR (this time is the
580 output of a SMP machine):
582 > cat /proc/interrupts
585 0: 1243498 1214548 IO-APIC-edge timer
586 1: 8949 8958 IO-APIC-edge keyboard
587 2: 0 0 XT-PIC cascade
588 5: 11286 10161 IO-APIC-edge soundblaster
589 8: 1 0 IO-APIC-edge rtc
590 9: 27422 27407 IO-APIC-edge 3c503
591 12: 113645 113873 IO-APIC-edge PS/2 Mouse
593 14: 22491 24012 IO-APIC-edge ide0
594 15: 2183 2415 IO-APIC-edge ide1
595 17: 30564 30414 IO-APIC-level eth0
596 18: 177 164 IO-APIC-level bttv
601 NMI is incremented in this case because every timer interrupt generates a NMI
602 (Non Maskable Interrupt) which is used by the NMI Watchdog to detect lockups.
604 LOC is the local interrupt counter of the internal APIC of every CPU.
606 ERR is incremented in the case of errors in the IO-APIC bus (the bus that
607 connects the CPUs in a SMP system. This means that an error has been detected,
608 the IO-APIC automatically retry the transmission, so it should not be a big
609 problem, but you should read the SMP-FAQ.
611 In 2.6.2* /proc/interrupts was expanded again. This time the goal was for
612 /proc/interrupts to display every IRQ vector in use by the system, not
613 just those considered 'most important'. The new vectors are:
615 THR -- interrupt raised when a machine check threshold counter
616 (typically counting ECC corrected errors of memory or cache) exceeds
617 a configurable threshold. Only available on some systems.
619 TRM -- a thermal event interrupt occurs when a temperature threshold
620 has been exceeded for the CPU. This interrupt may also be generated
621 when the temperature drops back to normal.
623 SPU -- a spurious interrupt is some interrupt that was raised then lowered
624 by some IO device before it could be fully processed by the APIC. Hence
625 the APIC sees the interrupt but does not know what device it came from.
626 For this case the APIC will generate the interrupt with a IRQ vector
627 of 0xff. This might also be generated by chipset bugs.
629 RES, CAL, TLB -- rescheduling, call and TLB flush interrupts are
630 sent from one CPU to another per the needs of the OS. Typically,
631 their statistics are used by kernel developers and interested users to
632 determine the occurrence of interrupts of the given type.
634 The above IRQ vectors are displayed only when relevant. For example,
635 the threshold vector does not exist on x86_64 platforms. Others are
636 suppressed when the system is a uniprocessor. As of this writing, only
637 i386 and x86_64 platforms support the new IRQ vector displays.
639 Of some interest is the introduction of the /proc/irq directory to 2.4.
640 It could be used to set IRQ to CPU affinity, this means that you can "hook" an
641 IRQ to only one CPU, or to exclude a CPU of handling IRQs. The contents of the
642 irq subdir is one subdir for each IRQ, and two files; default_smp_affinity and
647 0 10 12 14 16 18 2 4 6 8 prof_cpu_mask
648 1 11 13 15 17 19 3 5 7 9 default_smp_affinity
652 smp_affinity is a bitmask, in which you can specify which CPUs can handle the
653 IRQ, you can set it by doing:
655 > echo 1 > /proc/irq/10/smp_affinity
657 This means that only the first CPU will handle the IRQ, but you can also echo
658 5 which means that only the first and fourth CPU can handle the IRQ.
660 The contents of each smp_affinity file is the same by default:
662 > cat /proc/irq/0/smp_affinity
665 There is an alternate interface, smp_affinity_list which allows specifying
666 a cpu range instead of a bitmask:
668 > cat /proc/irq/0/smp_affinity_list
671 The default_smp_affinity mask applies to all non-active IRQs, which are the
672 IRQs which have not yet been allocated/activated, and hence which lack a
673 /proc/irq/[0-9]* directory.
675 The node file on an SMP system shows the node to which the device using the IRQ
676 reports itself as being attached. This hardware locality information does not
677 include information about any possible driver locality preference.
679 prof_cpu_mask specifies which CPUs are to be profiled by the system wide
680 profiler. Default value is ffffffff (all cpus if there are only 32 of them).
682 The way IRQs are routed is handled by the IO-APIC, and it's Round Robin
683 between all the CPUs which are allowed to handle it. As usual the kernel has
684 more info than you and does a better job than you, so the defaults are the
685 best choice for almost everyone. [Note this applies only to those IO-APIC's
686 that support "Round Robin" interrupt distribution.]
688 There are three more important subdirectories in /proc: net, scsi, and sys.
689 The general rule is that the contents, or even the existence of these
690 directories, depend on your kernel configuration. If SCSI is not enabled, the
691 directory scsi may not exist. The same is true with the net, which is there
692 only when networking support is present in the running kernel.
694 The slabinfo file gives information about memory usage at the slab level.
695 Linux uses slab pools for memory management above page level in version 2.2.
696 Commonly used objects have their own slab pool (such as network buffers,
697 directory cache, and so on).
699 ..............................................................................
701 > cat /proc/buddyinfo
703 Node 0, zone DMA 0 4 5 4 4 3 ...
704 Node 0, zone Normal 1 0 0 1 101 8 ...
705 Node 0, zone HighMem 2 0 0 1 1 0 ...
707 External fragmentation is a problem under some workloads, and buddyinfo is a
708 useful tool for helping diagnose these problems. Buddyinfo will give you a
709 clue as to how big an area you can safely allocate, or why a previous
712 Each column represents the number of pages of a certain order which are
713 available. In this case, there are 0 chunks of 2^0*PAGE_SIZE available in
714 ZONE_DMA, 4 chunks of 2^1*PAGE_SIZE in ZONE_DMA, 101 chunks of 2^4*PAGE_SIZE
715 available in ZONE_NORMAL, etc...
717 More information relevant to external fragmentation can be found in
720 > cat /proc/pagetypeinfo
724 Free pages count per migrate type at order 0 1 2 3 4 5 6 7 8 9 10
725 Node 0, zone DMA, type Unmovable 0 0 0 1 1 1 1 1 1 1 0
726 Node 0, zone DMA, type Reclaimable 0 0 0 0 0 0 0 0 0 0 0
727 Node 0, zone DMA, type Movable 1 1 2 1 2 1 1 0 1 0 2
728 Node 0, zone DMA, type Reserve 0 0 0 0 0 0 0 0 0 1 0
729 Node 0, zone DMA, type Isolate 0 0 0 0 0 0 0 0 0 0 0
730 Node 0, zone DMA32, type Unmovable 103 54 77 1 1 1 11 8 7 1 9
731 Node 0, zone DMA32, type Reclaimable 0 0 2 1 0 0 0 0 1 0 0
732 Node 0, zone DMA32, type Movable 169 152 113 91 77 54 39 13 6 1 452
733 Node 0, zone DMA32, type Reserve 1 2 2 2 2 0 1 1 1 1 0
734 Node 0, zone DMA32, type Isolate 0 0 0 0 0 0 0 0 0 0 0
736 Number of blocks type Unmovable Reclaimable Movable Reserve Isolate
737 Node 0, zone DMA 2 0 5 1 0
738 Node 0, zone DMA32 41 6 967 2 0
740 Fragmentation avoidance in the kernel works by grouping pages of different
741 migrate types into the same contiguous regions of memory called page blocks.
742 A page block is typically the size of the default hugepage size e.g. 2MB on
743 X86-64. By keeping pages grouped based on their ability to move, the kernel
744 can reclaim pages within a page block to satisfy a high-order allocation.
746 The pagetypinfo begins with information on the size of a page block. It
747 then gives the same type of information as buddyinfo except broken down
748 by migrate-type and finishes with details on how many page blocks of each
751 If min_free_kbytes has been tuned correctly (recommendations made by hugeadm
752 from libhugetlbfs http://sourceforge.net/projects/libhugetlbfs/), one can
753 make an estimate of the likely number of huge pages that can be allocated
754 at a given point in time. All the "Movable" blocks should be allocatable
755 unless memory has been mlock()'d. Some of the Reclaimable blocks should
756 also be allocatable although a lot of filesystem metadata may have to be
757 reclaimed to achieve this.
759 ..............................................................................
763 Provides information about distribution and utilization of memory. This
764 varies by architecture and compile options. The following is from a
765 16GB PIII, which has highmem enabled. You may not have all of these fields.
769 The "Locked" indicates whether the mapping is locked in memory or not.
772 MemTotal: 16344972 kB
779 HighTotal: 15597528 kB
780 HighFree: 13629632 kB
790 SReclaimable: 159856 kB
791 SUnreclaim: 124508 kB
796 CommitLimit: 7669796 kB
797 Committed_AS: 100056 kB
798 VmallocTotal: 112216 kB
800 VmallocChunk: 111088 kB
801 AnonHugePages: 49152 kB
803 MemTotal: Total usable ram (i.e. physical ram minus a few reserved
804 bits and the kernel binary code)
805 MemFree: The sum of LowFree+HighFree
806 Buffers: Relatively temporary storage for raw disk blocks
807 shouldn't get tremendously large (20MB or so)
808 Cached: in-memory cache for files read from the disk (the
809 pagecache). Doesn't include SwapCached
810 SwapCached: Memory that once was swapped out, is swapped back in but
811 still also is in the swapfile (if memory is needed it
812 doesn't need to be swapped out AGAIN because it is already
813 in the swapfile. This saves I/O)
814 Active: Memory that has been used more recently and usually not
815 reclaimed unless absolutely necessary.
816 Inactive: Memory which has been less recently used. It is more
817 eligible to be reclaimed for other purposes
819 HighFree: Highmem is all memory above ~860MB of physical memory
820 Highmem areas are for use by userspace programs, or
821 for the pagecache. The kernel must use tricks to access
822 this memory, making it slower to access than lowmem.
824 LowFree: Lowmem is memory which can be used for everything that
825 highmem can be used for, but it is also available for the
826 kernel's use for its own data structures. Among many
827 other things, it is where everything from the Slab is
828 allocated. Bad things happen when you're out of lowmem.
829 SwapTotal: total amount of swap space available
830 SwapFree: Memory which has been evicted from RAM, and is temporarily
832 Dirty: Memory which is waiting to get written back to the disk
833 Writeback: Memory which is actively being written back to the disk
834 AnonPages: Non-file backed pages mapped into userspace page tables
835 AnonHugePages: Non-file backed huge pages mapped into userspace page tables
836 Mapped: files which have been mmaped, such as libraries
837 Slab: in-kernel data structures cache
838 SReclaimable: Part of Slab, that might be reclaimed, such as caches
839 SUnreclaim: Part of Slab, that cannot be reclaimed on memory pressure
840 PageTables: amount of memory dedicated to the lowest level of page
842 NFS_Unstable: NFS pages sent to the server, but not yet committed to stable
844 Bounce: Memory used for block device "bounce buffers"
845 WritebackTmp: Memory used by FUSE for temporary writeback buffers
846 CommitLimit: Based on the overcommit ratio ('vm.overcommit_ratio'),
847 this is the total amount of memory currently available to
848 be allocated on the system. This limit is only adhered to
849 if strict overcommit accounting is enabled (mode 2 in
850 'vm.overcommit_memory').
851 The CommitLimit is calculated with the following formula:
852 CommitLimit = ('vm.overcommit_ratio' * Physical RAM) + Swap
853 For example, on a system with 1G of physical RAM and 7G
854 of swap with a `vm.overcommit_ratio` of 30 it would
855 yield a CommitLimit of 7.3G.
856 For more details, see the memory overcommit documentation
857 in vm/overcommit-accounting.
858 Committed_AS: The amount of memory presently allocated on the system.
859 The committed memory is a sum of all of the memory which
860 has been allocated by processes, even if it has not been
861 "used" by them as of yet. A process which malloc()'s 1G
862 of memory, but only touches 300M of it will only show up
863 as using 300M of memory even if it has the address space
864 allocated for the entire 1G. This 1G is memory which has
865 been "committed" to by the VM and can be used at any time
866 by the allocating application. With strict overcommit
867 enabled on the system (mode 2 in 'vm.overcommit_memory'),
868 allocations which would exceed the CommitLimit (detailed
869 above) will not be permitted. This is useful if one needs
870 to guarantee that processes will not fail due to lack of
871 memory once that memory has been successfully allocated.
872 VmallocTotal: total size of vmalloc memory area
873 VmallocUsed: amount of vmalloc area which is used
874 VmallocChunk: largest contiguous block of vmalloc area which is free
876 ..............................................................................
880 Provides information about vmalloced/vmaped areas. One line per area,
881 containing the virtual address range of the area, size in bytes,
882 caller information of the creator, and optional information depending
883 on the kind of area :
885 pages=nr number of pages
886 phys=addr if a physical address was specified
887 ioremap I/O mapping (ioremap() and friends)
888 vmalloc vmalloc() area
891 vpages buffer for pages pointers was vmalloced (huge area)
892 N<node>=nr (Only on NUMA kernels)
893 Number of pages allocated on memory node <node>
895 > cat /proc/vmallocinfo
896 0xffffc20000000000-0xffffc20000201000 2101248 alloc_large_system_hash+0x204 ...
897 /0x2c0 pages=512 vmalloc N0=128 N1=128 N2=128 N3=128
898 0xffffc20000201000-0xffffc20000302000 1052672 alloc_large_system_hash+0x204 ...
899 /0x2c0 pages=256 vmalloc N0=64 N1=64 N2=64 N3=64
900 0xffffc20000302000-0xffffc20000304000 8192 acpi_tb_verify_table+0x21/0x4f...
901 phys=7fee8000 ioremap
902 0xffffc20000304000-0xffffc20000307000 12288 acpi_tb_verify_table+0x21/0x4f...
903 phys=7fee7000 ioremap
904 0xffffc2000031d000-0xffffc2000031f000 8192 init_vdso_vars+0x112/0x210
905 0xffffc2000031f000-0xffffc2000032b000 49152 cramfs_uncompress_init+0x2e ...
906 /0x80 pages=11 vmalloc N0=3 N1=3 N2=2 N3=3
907 0xffffc2000033a000-0xffffc2000033d000 12288 sys_swapon+0x640/0xac0 ...
909 0xffffc20000347000-0xffffc2000034c000 20480 xt_alloc_table_info+0xfe ...
910 /0x130 [x_tables] pages=4 vmalloc N0=4
911 0xffffffffa0000000-0xffffffffa000f000 61440 sys_init_module+0xc27/0x1d00 ...
912 pages=14 vmalloc N2=14
913 0xffffffffa000f000-0xffffffffa0014000 20480 sys_init_module+0xc27/0x1d00 ...
915 0xffffffffa0014000-0xffffffffa0017000 12288 sys_init_module+0xc27/0x1d00 ...
917 0xffffffffa0017000-0xffffffffa0022000 45056 sys_init_module+0xc27/0x1d00 ...
918 pages=10 vmalloc N0=10
920 ..............................................................................
924 Provides counts of softirq handlers serviced since boot time, for each cpu.
929 TIMER: 27166 27120 27097 27034
934 SCHED: 27035 26983 26971 26746
936 RCU: 1678 1769 2178 2250
939 1.3 IDE devices in /proc/ide
940 ----------------------------
942 The subdirectory /proc/ide contains information about all IDE devices of which
943 the kernel is aware. There is one subdirectory for each IDE controller, the
944 file drivers and a link for each IDE device, pointing to the device directory
945 in the controller specific subtree.
947 The file drivers contains general information about the drivers used for the
950 > cat /proc/ide/drivers
951 ide-cdrom version 4.53
952 ide-disk version 1.08
954 More detailed information can be found in the controller specific
955 subdirectories. These are named ide0, ide1 and so on. Each of these
956 directories contains the files shown in table 1-6.
959 Table 1-6: IDE controller info in /proc/ide/ide?
960 ..............................................................................
962 channel IDE channel (0 or 1)
963 config Configuration (only for PCI/IDE bridge)
965 model Type/Chipset of IDE controller
966 ..............................................................................
968 Each device connected to a controller has a separate subdirectory in the
969 controllers directory. The files listed in table 1-7 are contained in these
973 Table 1-7: IDE device information
974 ..............................................................................
977 capacity Capacity of the medium (in 512Byte blocks)
978 driver driver and version
979 geometry physical and logical geometry
980 identify device identify block
982 model device identifier
983 settings device setup
984 smart_thresholds IDE disk management thresholds
985 smart_values IDE disk management values
986 ..............................................................................
988 The most interesting file is settings. This file contains a nice overview of
989 the drive parameters:
991 # cat /proc/ide/ide0/hda/settings
992 name value min max mode
993 ---- ----- --- --- ----
994 bios_cyl 526 0 65535 rw
995 bios_head 255 0 255 rw
997 breada_readahead 4 0 127 rw
999 file_readahead 72 0 2097151 rw
1001 keepsettings 0 0 1 rw
1002 max_kb_per_request 122 1 127 rw
1006 pio_mode write-only 0 255 w
1012 1.4 Networking info in /proc/net
1013 --------------------------------
1015 The subdirectory /proc/net follows the usual pattern. Table 1-8 shows the
1016 additional values you get for IP version 6 if you configure the kernel to
1017 support this. Table 1-9 lists the files and their meaning.
1020 Table 1-8: IPv6 info in /proc/net
1021 ..............................................................................
1023 udp6 UDP sockets (IPv6)
1024 tcp6 TCP sockets (IPv6)
1025 raw6 Raw device statistics (IPv6)
1026 igmp6 IP multicast addresses, which this host joined (IPv6)
1027 if_inet6 List of IPv6 interface addresses
1028 ipv6_route Kernel routing table for IPv6
1029 rt6_stats Global IPv6 routing tables statistics
1030 sockstat6 Socket statistics (IPv6)
1031 snmp6 Snmp data (IPv6)
1032 ..............................................................................
1035 Table 1-9: Network info in /proc/net
1036 ..............................................................................
1038 arp Kernel ARP table
1039 dev network devices with statistics
1040 dev_mcast the Layer2 multicast groups a device is listening too
1041 (interface index, label, number of references, number of bound
1043 dev_stat network device status
1044 ip_fwchains Firewall chain linkage
1045 ip_fwnames Firewall chain names
1046 ip_masq Directory containing the masquerading tables
1047 ip_masquerade Major masquerading table
1048 netstat Network statistics
1049 raw raw device statistics
1050 route Kernel routing table
1051 rpc Directory containing rpc info
1052 rt_cache Routing cache
1054 sockstat Socket statistics
1057 unix UNIX domain sockets
1058 wireless Wireless interface data (Wavelan etc)
1059 igmp IP multicast addresses, which this host joined
1060 psched Global packet scheduler parameters.
1061 netlink List of PF_NETLINK sockets
1062 ip_mr_vifs List of multicast virtual interfaces
1063 ip_mr_cache List of multicast routing cache
1064 ..............................................................................
1066 You can use this information to see which network devices are available in
1067 your system and how much traffic was routed over those devices:
1070 Inter-|Receive |[...
1071 face |bytes packets errs drop fifo frame compressed multicast|[...
1072 lo: 908188 5596 0 0 0 0 0 0 [...
1073 ppp0:15475140 20721 410 0 0 410 0 0 [...
1074 eth0: 614530 7085 0 0 0 0 0 1 [...
1077 ...] bytes packets errs drop fifo colls carrier compressed
1078 ...] 908188 5596 0 0 0 0 0 0
1079 ...] 1375103 17405 0 0 0 0 0 0
1080 ...] 1703981 5535 0 0 0 3 0 0
1082 In addition, each Channel Bond interface has its own directory. For
1083 example, the bond0 device will have a directory called /proc/net/bond0/.
1084 It will contain information that is specific to that bond, such as the
1085 current slaves of the bond, the link status of the slaves, and how
1086 many times the slaves link has failed.
1091 If you have a SCSI host adapter in your system, you'll find a subdirectory
1092 named after the driver for this adapter in /proc/scsi. You'll also see a list
1093 of all recognized SCSI devices in /proc/scsi:
1095 >cat /proc/scsi/scsi
1097 Host: scsi0 Channel: 00 Id: 00 Lun: 00
1098 Vendor: IBM Model: DGHS09U Rev: 03E0
1099 Type: Direct-Access ANSI SCSI revision: 03
1100 Host: scsi0 Channel: 00 Id: 06 Lun: 00
1101 Vendor: PIONEER Model: CD-ROM DR-U06S Rev: 1.04
1102 Type: CD-ROM ANSI SCSI revision: 02
1105 The directory named after the driver has one file for each adapter found in
1106 the system. These files contain information about the controller, including
1107 the used IRQ and the IO address range. The amount of information shown is
1108 dependent on the adapter you use. The example shows the output for an Adaptec
1109 AHA-2940 SCSI adapter:
1111 > cat /proc/scsi/aic7xxx/0
1113 Adaptec AIC7xxx driver version: 5.1.19/3.2.4
1115 TCQ Enabled By Default : Disabled
1116 AIC7XXX_PROC_STATS : Disabled
1117 AIC7XXX_RESET_DELAY : 5
1118 Adapter Configuration:
1119 SCSI Adapter: Adaptec AHA-294X Ultra SCSI host adapter
1120 Ultra Wide Controller
1121 PCI MMAPed I/O Base: 0xeb001000
1122 Adapter SEEPROM Config: SEEPROM found and used.
1123 Adaptec SCSI BIOS: Enabled
1125 SCBs: Active 0, Max Active 2,
1126 Allocated 15, HW 16, Page 255
1128 BIOS Control Word: 0x18b6
1129 Adapter Control Word: 0x005b
1130 Extended Translation: Enabled
1131 Disconnect Enable Flags: 0xffff
1132 Ultra Enable Flags: 0x0001
1133 Tag Queue Enable Flags: 0x0000
1134 Ordered Queue Tag Flags: 0x0000
1135 Default Tag Queue Depth: 8
1136 Tagged Queue By Device array for aic7xxx host instance 0:
1137 {255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255}
1138 Actual queue depth per device for aic7xxx host instance 0:
1139 {1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1}
1142 Device using Wide/Sync transfers at 40.0 MByte/sec, offset 8
1143 Transinfo settings: current(12/8/1/0), goal(12/8/1/0), user(12/15/1/0)
1144 Total transfers 160151 (74577 reads and 85574 writes)
1146 Device using Narrow/Sync transfers at 5.0 MByte/sec, offset 15
1147 Transinfo settings: current(50/15/0/0), goal(50/15/0/0), user(50/15/0/0)
1148 Total transfers 0 (0 reads and 0 writes)
1151 1.6 Parallel port info in /proc/parport
1152 ---------------------------------------
1154 The directory /proc/parport contains information about the parallel ports of
1155 your system. It has one subdirectory for each port, named after the port
1158 These directories contain the four files shown in Table 1-10.
1161 Table 1-10: Files in /proc/parport
1162 ..............................................................................
1164 autoprobe Any IEEE-1284 device ID information that has been acquired.
1165 devices list of the device drivers using that port. A + will appear by the
1166 name of the device currently using the port (it might not appear
1168 hardware Parallel port's base address, IRQ line and DMA channel.
1169 irq IRQ that parport is using for that port. This is in a separate
1170 file to allow you to alter it by writing a new value in (IRQ
1172 ..............................................................................
1174 1.7 TTY info in /proc/tty
1175 -------------------------
1177 Information about the available and actually used tty's can be found in the
1178 directory /proc/tty.You'll find entries for drivers and line disciplines in
1179 this directory, as shown in Table 1-11.
1182 Table 1-11: Files in /proc/tty
1183 ..............................................................................
1185 drivers list of drivers and their usage
1186 ldiscs registered line disciplines
1187 driver/serial usage statistic and status of single tty lines
1188 ..............................................................................
1190 To see which tty's are currently in use, you can simply look into the file
1193 > cat /proc/tty/drivers
1194 pty_slave /dev/pts 136 0-255 pty:slave
1195 pty_master /dev/ptm 128 0-255 pty:master
1196 pty_slave /dev/ttyp 3 0-255 pty:slave
1197 pty_master /dev/pty 2 0-255 pty:master
1198 serial /dev/cua 5 64-67 serial:callout
1199 serial /dev/ttyS 4 64-67 serial
1200 /dev/tty0 /dev/tty0 4 0 system:vtmaster
1201 /dev/ptmx /dev/ptmx 5 2 system
1202 /dev/console /dev/console 5 1 system:console
1203 /dev/tty /dev/tty 5 0 system:/dev/tty
1204 unknown /dev/tty 4 1-63 console
1207 1.8 Miscellaneous kernel statistics in /proc/stat
1208 -------------------------------------------------
1210 Various pieces of information about kernel activity are available in the
1211 /proc/stat file. All of the numbers reported in this file are aggregates
1212 since the system first booted. For a quick look, simply cat the file:
1215 cpu 2255 34 2290 22625563 6290 127 456 0 0
1216 cpu0 1132 34 1441 11311718 3675 127 438 0 0
1217 cpu1 1123 0 849 11313845 2614 0 18 0 0
1218 intr 114930548 113199788 3 0 5 263 0 4 [... lots more numbers ...]
1224 softirq 183433 0 21755 12 39 1137 231 21459 2263
1226 The very first "cpu" line aggregates the numbers in all of the other "cpuN"
1227 lines. These numbers identify the amount of time the CPU has spent performing
1228 different kinds of work. Time units are in USER_HZ (typically hundredths of a
1229 second). The meanings of the columns are as follows, from left to right:
1231 - user: normal processes executing in user mode
1232 - nice: niced processes executing in user mode
1233 - system: processes executing in kernel mode
1234 - idle: twiddling thumbs
1235 - iowait: waiting for I/O to complete
1236 - irq: servicing interrupts
1237 - softirq: servicing softirqs
1238 - steal: involuntary wait
1239 - guest: running a normal guest
1240 - guest_nice: running a niced guest
1242 The "intr" line gives counts of interrupts serviced since boot time, for each
1243 of the possible system interrupts. The first column is the total of all
1244 interrupts serviced; each subsequent column is the total for that particular
1247 The "ctxt" line gives the total number of context switches across all CPUs.
1249 The "btime" line gives the time at which the system booted, in seconds since
1252 The "processes" line gives the number of processes and threads created, which
1253 includes (but is not limited to) those created by calls to the fork() and
1254 clone() system calls.
1256 The "procs_running" line gives the total number of threads that are
1257 running or ready to run (i.e., the total number of runnable threads).
1259 The "procs_blocked" line gives the number of processes currently blocked,
1260 waiting for I/O to complete.
1262 The "softirq" line gives counts of softirqs serviced since boot time, for each
1263 of the possible system softirqs. The first column is the total of all
1264 softirqs serviced; each subsequent column is the total for that particular
1268 1.9 Ext4 file system parameters
1269 ------------------------------
1271 Information about mounted ext4 file systems can be found in
1272 /proc/fs/ext4. Each mounted filesystem will have a directory in
1273 /proc/fs/ext4 based on its device name (i.e., /proc/fs/ext4/hdc or
1274 /proc/fs/ext4/dm-0). The files in each per-device directory are shown
1275 in Table 1-12, below.
1277 Table 1-12: Files in /proc/fs/ext4/<devname>
1278 ..............................................................................
1280 mb_groups details of multiblock allocator buddy cache of free blocks
1281 ..............................................................................
1285 Shows registered system console lines.
1287 To see which character device lines are currently used for the system console
1288 /dev/console, you may simply look into the file /proc/consoles:
1290 > cat /proc/consoles
1296 device name of the device
1297 operations R = can do read operations
1298 W = can do write operations
1300 flags E = it is enabled
1301 C = it is preferred console
1302 B = it is primary boot console
1303 p = it is used for printk buffer
1304 b = it is not a TTY but a Braille device
1305 a = it is safe to use when cpu is offline
1306 major:minor major and minor number of the device separated by a colon
1308 ------------------------------------------------------------------------------
1310 ------------------------------------------------------------------------------
1311 The /proc file system serves information about the running system. It not only
1312 allows access to process data but also allows you to request the kernel status
1313 by reading files in the hierarchy.
1315 The directory structure of /proc reflects the types of information and makes
1316 it easy, if not obvious, where to look for specific data.
1317 ------------------------------------------------------------------------------
1319 ------------------------------------------------------------------------------
1320 CHAPTER 2: MODIFYING SYSTEM PARAMETERS
1321 ------------------------------------------------------------------------------
1323 ------------------------------------------------------------------------------
1325 ------------------------------------------------------------------------------
1326 * Modifying kernel parameters by writing into files found in /proc/sys
1327 * Exploring the files which modify certain parameters
1328 * Review of the /proc/sys file tree
1329 ------------------------------------------------------------------------------
1332 A very interesting part of /proc is the directory /proc/sys. This is not only
1333 a source of information, it also allows you to change parameters within the
1334 kernel. Be very careful when attempting this. You can optimize your system,
1335 but you can also cause it to crash. Never alter kernel parameters on a
1336 production system. Set up a development machine and test to make sure that
1337 everything works the way you want it to. You may have no alternative but to
1338 reboot the machine once an error has been made.
1340 To change a value, simply echo the new value into the file. An example is
1341 given below in the section on the file system data. You need to be root to do
1342 this. You can create your own boot script to perform this every time your
1345 The files in /proc/sys can be used to fine tune and monitor miscellaneous and
1346 general things in the operation of the Linux kernel. Since some of the files
1347 can inadvertently disrupt your system, it is advisable to read both
1348 documentation and source before actually making adjustments. In any case, be
1349 very careful when writing to any of these files. The entries in /proc may
1350 change slightly between the 2.1.* and the 2.2 kernel, so if there is any doubt
1351 review the kernel documentation in the directory /usr/src/linux/Documentation.
1352 This chapter is heavily based on the documentation included in the pre 2.2
1353 kernels, and became part of it in version 2.2.1 of the Linux kernel.
1355 Please see: Documentation/sysctl/ directory for descriptions of these
1358 ------------------------------------------------------------------------------
1360 ------------------------------------------------------------------------------
1361 Certain aspects of kernel behavior can be modified at runtime, without the
1362 need to recompile the kernel, or even to reboot the system. The files in the
1363 /proc/sys tree can not only be read, but also modified. You can use the echo
1364 command to write value into these files, thereby changing the default settings
1366 ------------------------------------------------------------------------------
1368 ------------------------------------------------------------------------------
1369 CHAPTER 3: PER-PROCESS PARAMETERS
1370 ------------------------------------------------------------------------------
1372 3.1 /proc/<pid>/oom_adj & /proc/<pid>/oom_score_adj- Adjust the oom-killer score
1373 --------------------------------------------------------------------------------
1375 These file can be used to adjust the badness heuristic used to select which
1376 process gets killed in out of memory conditions.
1378 The badness heuristic assigns a value to each candidate task ranging from 0
1379 (never kill) to 1000 (always kill) to determine which process is targeted. The
1380 units are roughly a proportion along that range of allowed memory the process
1381 may allocate from based on an estimation of its current memory and swap use.
1382 For example, if a task is using all allowed memory, its badness score will be
1383 1000. If it is using half of its allowed memory, its score will be 500.
1385 There is an additional factor included in the badness score: the current memory
1386 and swap usage is discounted by 3% for root processes.
1388 The amount of "allowed" memory depends on the context in which the oom killer
1389 was called. If it is due to the memory assigned to the allocating task's cpuset
1390 being exhausted, the allowed memory represents the set of mems assigned to that
1391 cpuset. If it is due to a mempolicy's node(s) being exhausted, the allowed
1392 memory represents the set of mempolicy nodes. If it is due to a memory
1393 limit (or swap limit) being reached, the allowed memory is that configured
1394 limit. Finally, if it is due to the entire system being out of memory, the
1395 allowed memory represents all allocatable resources.
1397 The value of /proc/<pid>/oom_score_adj is added to the badness score before it
1398 is used to determine which task to kill. Acceptable values range from -1000
1399 (OOM_SCORE_ADJ_MIN) to +1000 (OOM_SCORE_ADJ_MAX). This allows userspace to
1400 polarize the preference for oom killing either by always preferring a certain
1401 task or completely disabling it. The lowest possible value, -1000, is
1402 equivalent to disabling oom killing entirely for that task since it will always
1403 report a badness score of 0.
1405 Consequently, it is very simple for userspace to define the amount of memory to
1406 consider for each task. Setting a /proc/<pid>/oom_score_adj value of +500, for
1407 example, is roughly equivalent to allowing the remainder of tasks sharing the
1408 same system, cpuset, mempolicy, or memory controller resources to use at least
1409 50% more memory. A value of -500, on the other hand, would be roughly
1410 equivalent to discounting 50% of the task's allowed memory from being considered
1411 as scoring against the task.
1413 For backwards compatibility with previous kernels, /proc/<pid>/oom_adj may also
1414 be used to tune the badness score. Its acceptable values range from -16
1415 (OOM_ADJUST_MIN) to +15 (OOM_ADJUST_MAX) and a special value of -17
1416 (OOM_DISABLE) to disable oom killing entirely for that task. Its value is
1417 scaled linearly with /proc/<pid>/oom_score_adj.
1419 The value of /proc/<pid>/oom_score_adj may be reduced no lower than the last
1420 value set by a CAP_SYS_RESOURCE process. To reduce the value any lower
1421 requires CAP_SYS_RESOURCE.
1423 Caveat: when a parent task is selected, the oom killer will sacrifice any first
1424 generation children with separate address spaces instead, if possible. This
1425 avoids servers and important system daemons from being killed and loses the
1426 minimal amount of work.
1429 3.2 /proc/<pid>/oom_score - Display current oom-killer score
1430 -------------------------------------------------------------
1432 This file can be used to check the current score used by the oom-killer is for
1433 any given <pid>. Use it together with /proc/<pid>/oom_score_adj to tune which
1434 process should be killed in an out-of-memory situation.
1437 3.3 /proc/<pid>/io - Display the IO accounting fields
1438 -------------------------------------------------------
1440 This file contains IO statistics for each running process
1445 test:/tmp # dd if=/dev/zero of=/tmp/test.dat &
1448 test:/tmp # cat /proc/3828/io
1454 write_bytes: 323932160
1455 cancelled_write_bytes: 0
1464 I/O counter: chars read
1465 The number of bytes which this task has caused to be read from storage. This
1466 is simply the sum of bytes which this process passed to read() and pread().
1467 It includes things like tty IO and it is unaffected by whether or not actual
1468 physical disk IO was required (the read might have been satisfied from
1475 I/O counter: chars written
1476 The number of bytes which this task has caused, or shall cause to be written
1477 to disk. Similar caveats apply here as with rchar.
1483 I/O counter: read syscalls
1484 Attempt to count the number of read I/O operations, i.e. syscalls like read()
1491 I/O counter: write syscalls
1492 Attempt to count the number of write I/O operations, i.e. syscalls like
1493 write() and pwrite().
1499 I/O counter: bytes read
1500 Attempt to count the number of bytes which this process really did cause to
1501 be fetched from the storage layer. Done at the submit_bio() level, so it is
1502 accurate for block-backed filesystems. <please add status regarding NFS and
1503 CIFS at a later time>
1509 I/O counter: bytes written
1510 Attempt to count the number of bytes which this process caused to be sent to
1511 the storage layer. This is done at page-dirtying time.
1514 cancelled_write_bytes
1515 ---------------------
1517 The big inaccuracy here is truncate. If a process writes 1MB to a file and
1518 then deletes the file, it will in fact perform no writeout. But it will have
1519 been accounted as having caused 1MB of write.
1520 In other words: The number of bytes which this process caused to not happen,
1521 by truncating pagecache. A task can cause "negative" IO too. If this task
1522 truncates some dirty pagecache, some IO which another task has been accounted
1523 for (in its write_bytes) will not be happening. We _could_ just subtract that
1524 from the truncating task's write_bytes, but there is information loss in doing
1531 At its current implementation state, this is a bit racy on 32-bit machines: if
1532 process A reads process B's /proc/pid/io while process B is updating one of
1533 those 64-bit counters, process A could see an intermediate result.
1536 More information about this can be found within the taskstats documentation in
1537 Documentation/accounting.
1539 3.4 /proc/<pid>/coredump_filter - Core dump filtering settings
1540 ---------------------------------------------------------------
1541 When a process is dumped, all anonymous memory is written to a core file as
1542 long as the size of the core file isn't limited. But sometimes we don't want
1543 to dump some memory segments, for example, huge shared memory. Conversely,
1544 sometimes we want to save file-backed memory segments into a core file, not
1545 only the individual files.
1547 /proc/<pid>/coredump_filter allows you to customize which memory segments
1548 will be dumped when the <pid> process is dumped. coredump_filter is a bitmask
1549 of memory types. If a bit of the bitmask is set, memory segments of the
1550 corresponding memory type are dumped, otherwise they are not dumped.
1552 The following 7 memory types are supported:
1553 - (bit 0) anonymous private memory
1554 - (bit 1) anonymous shared memory
1555 - (bit 2) file-backed private memory
1556 - (bit 3) file-backed shared memory
1557 - (bit 4) ELF header pages in file-backed private memory areas (it is
1558 effective only if the bit 2 is cleared)
1559 - (bit 5) hugetlb private memory
1560 - (bit 6) hugetlb shared memory
1562 Note that MMIO pages such as frame buffer are never dumped and vDSO pages
1563 are always dumped regardless of the bitmask status.
1565 Note bit 0-4 doesn't effect any hugetlb memory. hugetlb memory are only
1566 effected by bit 5-6.
1568 Default value of coredump_filter is 0x23; this means all anonymous memory
1569 segments and hugetlb private memory are dumped.
1571 If you don't want to dump all shared memory segments attached to pid 1234,
1572 write 0x21 to the process's proc file.
1574 $ echo 0x21 > /proc/1234/coredump_filter
1576 When a new process is created, the process inherits the bitmask status from its
1577 parent. It is useful to set up coredump_filter before the program runs.
1580 $ echo 0x7 > /proc/self/coredump_filter
1583 3.5 /proc/<pid>/mountinfo - Information about mounts
1584 --------------------------------------------------------
1586 This file contains lines of the form:
1588 36 35 98:0 /mnt1 /mnt2 rw,noatime master:1 - ext3 /dev/root rw,errors=continue
1589 (1)(2)(3) (4) (5) (6) (7) (8) (9) (10) (11)
1591 (1) mount ID: unique identifier of the mount (may be reused after umount)
1592 (2) parent ID: ID of parent (or of self for the top of the mount tree)
1593 (3) major:minor: value of st_dev for files on filesystem
1594 (4) root: root of the mount within the filesystem
1595 (5) mount point: mount point relative to the process's root
1596 (6) mount options: per mount options
1597 (7) optional fields: zero or more fields of the form "tag[:value]"
1598 (8) separator: marks the end of the optional fields
1599 (9) filesystem type: name of filesystem of the form "type[.subtype]"
1600 (10) mount source: filesystem specific information or "none"
1601 (11) super options: per super block options
1603 Parsers should ignore all unrecognised optional fields. Currently the
1604 possible optional fields are:
1606 shared:X mount is shared in peer group X
1607 master:X mount is slave to peer group X
1608 propagate_from:X mount is slave and receives propagation from peer group X (*)
1609 unbindable mount is unbindable
1611 (*) X is the closest dominant peer group under the process's root. If
1612 X is the immediate master of the mount, or if there's no dominant peer
1613 group under the same root, then only the "master:X" field is present
1614 and not the "propagate_from:X" field.
1616 For more information on mount propagation see:
1618 Documentation/filesystems/sharedsubtree.txt
1621 3.6 /proc/<pid>/comm & /proc/<pid>/task/<tid>/comm
1622 --------------------------------------------------------
1623 These files provide a method to access a tasks comm value. It also allows for
1624 a task to set its own or one of its thread siblings comm value. The comm value
1625 is limited in size compared to the cmdline value, so writing anything longer
1626 then the kernel's TASK_COMM_LEN (currently 16 chars) will result in a truncated
1630 3.7 /proc/<pid>/task/<tid>/children - Information about task children
1631 -------------------------------------------------------------------------
1632 This file provides a fast way to retrieve first level children pids
1633 of a task pointed by <pid>/<tid> pair. The format is a space separated
1636 Note the "first level" here -- if a child has own children they will
1637 not be listed here, one needs to read /proc/<children-pid>/task/<tid>/children
1638 to obtain the descendants.
1640 Since this interface is intended to be fast and cheap it doesn't
1641 guarantee to provide precise results and some children might be
1642 skipped, especially if they've exited right after we printed their
1643 pids, so one need to either stop or freeze processes being inspected
1644 if precise results are needed.
1647 3.7 /proc/<pid>/fdinfo/<fd> - Information about opened file
1648 ---------------------------------------------------------------
1649 This file provides information associated with an opened file. The regular
1650 files have at least two fields -- 'pos' and 'flags'. The 'pos' represents
1651 the current offset of the opened file in decimal form [see lseek(2) for
1652 details] and 'flags' denotes the octal O_xxx mask the file has been
1653 created with [see open(2) for details].
1660 The files such as eventfd, fsnotify, signalfd, epoll among the regular pos/flags
1661 pair provide additional information particular to the objects they represent.
1669 where 'eventfd-count' is hex value of a counter.
1675 sigmask: 0000000000000200
1677 where 'sigmask' is hex value of the signal mask associated
1684 tfd: 5 events: 1d data: ffffffffffffffff
1686 where 'tfd' is a target file descriptor number in decimal form,
1687 'events' is events mask being watched and the 'data' is data
1688 associated with a target [see epoll(7) for more details].
1692 For inotify files the format is the following
1696 inotify wd:3 ino:9e7e sdev:800013 mask:800afce ignored_mask:0 fhandle-bytes:8 fhandle-type:1 f_handle:7e9e0000640d1b6d
1698 where 'wd' is a watch descriptor in decimal form, ie a target file
1699 descriptor number, 'ino' and 'sdev' are inode and device where the
1700 target file resides and the 'mask' is the mask of events, all in hex
1701 form [see inotify(7) for more details].
1703 If the kernel was built with exportfs support, the path to the target
1704 file is encoded as a file handle. The file handle is provided by three
1705 fields 'fhandle-bytes', 'fhandle-type' and 'f_handle', all in hex
1708 If the kernel is built without exportfs support the file handle won't be
1711 If there is no inotify mark attached yet the 'inotify' line will be omitted.
1713 For fanotify files the format is
1717 fanotify flags:10 event-flags:0
1718 fanotify mnt_id:12 mflags:40 mask:38 ignored_mask:40000003
1719 fanotify ino:4f969 sdev:800013 mflags:0 mask:3b ignored_mask:40000000 fhandle-bytes:8 fhandle-type:1 f_handle:69f90400c275b5b4
1721 where fanotify 'flags' and 'event-flags' are values used in fanotify_init
1722 call, 'mnt_id' is the mount point identifier, 'mflags' is the value of
1723 flags associated with mark which are tracked separately from events
1724 mask. 'ino', 'sdev' are target inode and device, 'mask' is the events
1725 mask and 'ignored_mask' is the mask of events which are to be ignored.
1726 All in hex format. Incorporation of 'mflags', 'mask' and 'ignored_mask'
1727 does provide information about flags and mask used in fanotify_mark
1728 call [see fsnotify manpage for details].
1730 While the first three lines are mandatory and always printed, the rest is
1731 optional and may be omitted if no marks created yet.
1734 ------------------------------------------------------------------------------
1736 ------------------------------------------------------------------------------
1739 ---------------------
1741 The following mount options are supported:
1743 hidepid= Set /proc/<pid>/ access mode.
1744 gid= Set the group authorized to learn processes information.
1746 hidepid=0 means classic mode - everybody may access all /proc/<pid>/ directories
1749 hidepid=1 means users may not access any /proc/<pid>/ directories but their
1750 own. Sensitive files like cmdline, sched*, status are now protected against
1751 other users. This makes it impossible to learn whether any user runs
1752 specific program (given the program doesn't reveal itself by its behaviour).
1753 As an additional bonus, as /proc/<pid>/cmdline is unaccessible for other users,
1754 poorly written programs passing sensitive information via program arguments are
1755 now protected against local eavesdroppers.
1757 hidepid=2 means hidepid=1 plus all /proc/<pid>/ will be fully invisible to other
1758 users. It doesn't mean that it hides a fact whether a process with a specific
1759 pid value exists (it can be learned by other means, e.g. by "kill -0 $PID"),
1760 but it hides process' uid and gid, which may be learned by stat()'ing
1761 /proc/<pid>/ otherwise. It greatly complicates an intruder's task of gathering
1762 information about running processes, whether some daemon runs with elevated
1763 privileges, whether other user runs some sensitive program, whether other users
1764 run any program at all, etc.
1766 gid= defines a group authorized to learn processes information otherwise
1767 prohibited by hidepid=. If you use some daemon like identd which needs to learn
1768 information about processes information, just add identd to this group.