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
45 ------------------------------------------------------------------------------
47 ------------------------------------------------------------------------------
49 0.1 Introduction/Credits
50 ------------------------
52 This documentation is part of a soon (or so we hope) to be released book on
53 the SuSE Linux distribution. As there is no complete documentation for the
54 /proc file system and we've used many freely available sources to write these
55 chapters, it seems only fair to give the work back to the Linux community.
56 This work is based on the 2.2.* kernel version and the upcoming 2.4.*. I'm
57 afraid it's still far from complete, but we hope it will be useful. As far as
58 we know, it is the first 'all-in-one' document about the /proc file system. It
59 is focused on the Intel x86 hardware, so if you are looking for PPC, ARM,
60 SPARC, AXP, etc., features, you probably won't find what you are looking for.
61 It also only covers IPv4 networking, not IPv6 nor other protocols - sorry. But
62 additions and patches are welcome and will be added to this document if you
65 We'd like to thank Alan Cox, Rik van Riel, and Alexey Kuznetsov and a lot of
66 other people for help compiling this documentation. We'd also like to extend a
67 special thank you to Andi Kleen for documentation, which we relied on heavily
68 to create this document, as well as the additional information he provided.
69 Thanks to everybody else who contributed source or docs to the Linux kernel
70 and helped create a great piece of software... :)
72 If you have any comments, corrections or additions, please don't hesitate to
73 contact Bodo Bauer at bb@ricochet.net. We'll be happy to add them to this
76 The latest version of this document is available online at
77 http://tldp.org/LDP/Linux-Filesystem-Hierarchy/html/proc.html
79 If the above direction does not works for you, you could try the kernel
80 mailing list at linux-kernel@vger.kernel.org and/or try to reach me at
81 comandante@zaralinux.com.
86 We don't guarantee the correctness of this document, and if you come to us
87 complaining about how you screwed up your system because of incorrect
88 documentation, we won't feel responsible...
90 ------------------------------------------------------------------------------
91 CHAPTER 1: COLLECTING SYSTEM INFORMATION
92 ------------------------------------------------------------------------------
94 ------------------------------------------------------------------------------
96 ------------------------------------------------------------------------------
97 * Investigating the properties of the pseudo file system /proc and its
98 ability to provide information on the running Linux system
99 * Examining /proc's structure
100 * Uncovering various information about the kernel and the processes running
102 ------------------------------------------------------------------------------
105 The proc file system acts as an interface to internal data structures in the
106 kernel. It can be used to obtain information about the system and to change
107 certain kernel parameters at runtime (sysctl).
109 First, we'll take a look at the read-only parts of /proc. In Chapter 2, we
110 show you how you can use /proc/sys to change settings.
112 1.1 Process-Specific Subdirectories
113 -----------------------------------
115 The directory /proc contains (among other things) one subdirectory for each
116 process running on the system, which is named after the process ID (PID).
118 The link self points to the process reading the file system. Each process
119 subdirectory has the entries listed in Table 1-1.
122 Table 1-1: Process specific entries in /proc
123 ..............................................................................
125 clear_refs Clears page referenced bits shown in smaps output
126 cmdline Command line arguments
127 cpu Current and last cpu in which it was executed (2.4)(smp)
128 cwd Link to the current working directory
129 environ Values of environment variables
130 exe Link to the executable of this process
131 fd Directory, which contains all file descriptors
132 maps Memory maps to executables and library files (2.4)
133 mem Memory held by this process
134 root Link to the root directory of this process
136 statm Process memory status information
137 status Process status in human readable form
138 wchan If CONFIG_KALLSYMS is set, a pre-decoded wchan
140 stack Report full stack trace, enable via CONFIG_STACKTRACE
141 smaps a extension based on maps, showing the memory consumption of
143 ..............................................................................
145 For example, to get the status information of a process, all you have to do is
146 read the file /proc/PID/status:
148 >cat /proc/self/status
172 SigPnd: 0000000000000000
173 ShdPnd: 0000000000000000
174 SigBlk: 0000000000000000
175 SigIgn: 0000000000000000
176 SigCgt: 0000000000000000
177 CapInh: 00000000fffffeff
178 CapPrm: 0000000000000000
179 CapEff: 0000000000000000
180 CapBnd: ffffffffffffffff
181 voluntary_ctxt_switches: 0
182 nonvoluntary_ctxt_switches: 1
184 This shows you nearly the same information you would get if you viewed it with
185 the ps command. In fact, ps uses the proc file system to obtain its
186 information. But you get a more detailed view of the process by reading the
187 file /proc/PID/status. It fields are described in table 1-2.
189 The statm file contains more detailed information about the process
190 memory usage. Its seven fields are explained in Table 1-3. The stat file
191 contains details information about the process itself. Its fields are
192 explained in Table 1-4.
194 (for SMP CONFIG users)
195 For making accounting scalable, RSS related information are handled in
196 asynchronous manner and the vaule may not be very precise. To see a precise
197 snapshot of a moment, you can see /proc/<pid>/smaps file and scan page table.
198 It's slow but very precise.
200 Table 1-2: Contents of the status files (as of 2.6.30-rc7)
201 ..............................................................................
203 Name filename of the executable
204 State state (R is running, S is sleeping, D is sleeping
205 in an uninterruptible wait, Z is zombie,
206 T is traced or stopped)
209 PPid process id of the parent process
210 TracerPid PID of process tracing this process (0 if not)
211 Uid Real, effective, saved set, and file system UIDs
212 Gid Real, effective, saved set, and file system GIDs
213 FDSize number of file descriptor slots currently allocated
214 Groups supplementary group list
215 VmPeak peak virtual memory size
216 VmSize total program size
217 VmLck locked memory size
218 VmHWM peak resident set size ("high water mark")
219 VmRSS size of memory portions
220 VmData size of data, stack, and text segments
221 VmStk size of data, stack, and text segments
222 VmExe size of text segment
223 VmLib size of shared library code
224 VmPTE size of page table entries
225 VmSwap size of swap usage (the number of referred swapents)
226 Threads number of threads
227 SigQ number of signals queued/max. number for queue
228 SigPnd bitmap of pending signals for the thread
229 ShdPnd bitmap of shared pending signals for the process
230 SigBlk bitmap of blocked signals
231 SigIgn bitmap of ignored signals
232 SigCgt bitmap of catched signals
233 CapInh bitmap of inheritable capabilities
234 CapPrm bitmap of permitted capabilities
235 CapEff bitmap of effective capabilities
236 CapBnd bitmap of capabilities bounding set
237 Cpus_allowed mask of CPUs on which this process may run
238 Cpus_allowed_list Same as previous, but in "list format"
239 Mems_allowed mask of memory nodes allowed to this process
240 Mems_allowed_list Same as previous, but in "list format"
241 voluntary_ctxt_switches number of voluntary context switches
242 nonvoluntary_ctxt_switches number of non voluntary context switches
243 ..............................................................................
245 Table 1-3: Contents of the statm files (as of 2.6.8-rc3)
246 ..............................................................................
248 size total program size (pages) (same as VmSize in status)
249 resident size of memory portions (pages) (same as VmRSS in status)
250 shared number of pages that are shared (i.e. backed by a file)
251 trs number of pages that are 'code' (not including libs; broken,
252 includes data segment)
253 lrs number of pages of library (always 0 on 2.6)
254 drs number of pages of data/stack (including libs; broken,
255 includes library text)
256 dt number of dirty pages (always 0 on 2.6)
257 ..............................................................................
260 Table 1-4: Contents of the stat files (as of 2.6.30-rc7)
261 ..............................................................................
264 tcomm filename of the executable
265 state state (R is running, S is sleeping, D is sleeping in an
266 uninterruptible wait, Z is zombie, T is traced or stopped)
267 ppid process id of the parent process
268 pgrp pgrp of the process
270 tty_nr tty the process uses
271 tty_pgrp pgrp of the tty
273 min_flt number of minor faults
274 cmin_flt number of minor faults with child's
275 maj_flt number of major faults
276 cmaj_flt number of major faults with child's
277 utime user mode jiffies
278 stime kernel mode jiffies
279 cutime user mode jiffies with child's
280 cstime kernel mode jiffies with child's
281 priority priority level
283 num_threads number of threads
284 it_real_value (obsolete, always 0)
285 start_time time the process started after system boot
286 vsize virtual memory size
287 rss resident set memory size
288 rsslim current limit in bytes on the rss
289 start_code address above which program text can run
290 end_code address below which program text can run
291 start_stack address of the start of the stack
292 esp current value of ESP
293 eip current value of EIP
294 pending bitmap of pending signals
295 blocked bitmap of blocked signals
296 sigign bitmap of ignored signals
297 sigcatch bitmap of catched signals
298 wchan address where process went to sleep
301 exit_signal signal to send to parent thread on exit
302 task_cpu which CPU the task is scheduled on
303 rt_priority realtime priority
304 policy scheduling policy (man sched_setscheduler)
305 blkio_ticks time spent waiting for block IO
306 gtime guest time of the task in jiffies
307 cgtime guest time of the task children in jiffies
308 ..............................................................................
310 The /proc/PID/maps file containing the currently mapped memory regions and
311 their access permissions.
315 address perms offset dev inode pathname
317 08048000-08049000 r-xp 00000000 03:00 8312 /opt/test
318 08049000-0804a000 rw-p 00001000 03:00 8312 /opt/test
319 0804a000-0806b000 rw-p 00000000 00:00 0 [heap]
320 a7cb1000-a7cb2000 ---p 00000000 00:00 0
321 a7cb2000-a7eb2000 rw-p 00000000 00:00 0
322 a7eb2000-a7eb3000 ---p 00000000 00:00 0
323 a7eb3000-a7ed5000 rw-p 00000000 00:00 0
324 a7ed5000-a8008000 r-xp 00000000 03:00 4222 /lib/libc.so.6
325 a8008000-a800a000 r--p 00133000 03:00 4222 /lib/libc.so.6
326 a800a000-a800b000 rw-p 00135000 03:00 4222 /lib/libc.so.6
327 a800b000-a800e000 rw-p 00000000 00:00 0
328 a800e000-a8022000 r-xp 00000000 03:00 14462 /lib/libpthread.so.0
329 a8022000-a8023000 r--p 00013000 03:00 14462 /lib/libpthread.so.0
330 a8023000-a8024000 rw-p 00014000 03:00 14462 /lib/libpthread.so.0
331 a8024000-a8027000 rw-p 00000000 00:00 0
332 a8027000-a8043000 r-xp 00000000 03:00 8317 /lib/ld-linux.so.2
333 a8043000-a8044000 r--p 0001b000 03:00 8317 /lib/ld-linux.so.2
334 a8044000-a8045000 rw-p 0001c000 03:00 8317 /lib/ld-linux.so.2
335 aff35000-aff4a000 rw-p 00000000 00:00 0 [stack]
336 ffffe000-fffff000 r-xp 00000000 00:00 0 [vdso]
338 where "address" is the address space in the process that it occupies, "perms"
339 is a set of permissions:
345 p = private (copy on write)
347 "offset" is the offset into the mapping, "dev" is the device (major:minor), and
348 "inode" is the inode on that device. 0 indicates that no inode is associated
349 with the memory region, as the case would be with BSS (uninitialized data).
350 The "pathname" shows the name associated file for this mapping. If the mapping
351 is not associated with a file:
353 [heap] = the heap of the program
354 [stack] = the stack of the main process
355 [vdso] = the "virtual dynamic shared object",
356 the kernel system call handler
358 or if empty, the mapping is anonymous.
361 The /proc/PID/smaps is an extension based on maps, showing the memory
362 consumption for each of the process's mappings. For each of mappings there
363 is a series of lines such as the following:
365 08048000-080bc000 r-xp 00000000 03:02 13130 /bin/bash
380 The first of these lines shows the same information as is displayed for the
381 mapping in /proc/PID/maps. The remaining lines show the size of the mapping
382 (size), the amount of the mapping that is currently resident in RAM (RSS), the
383 process' proportional share of this mapping (PSS), the number of clean and
384 dirty private pages in the mapping. Note that even a page which is part of a
385 MAP_SHARED mapping, but has only a single pte mapped, i.e. is currently used
386 by only one process, is accounted as private and not as shared. "Referenced"
387 indicates the amount of memory currently marked as referenced or accessed.
388 "Anonymous" shows the amount of memory that does not belong to any file. Even
389 a mapping associated with a file may contain anonymous pages: when MAP_PRIVATE
390 and a page is modified, the file page is replaced by a private anonymous copy.
391 "Swap" shows how much would-be-anonymous memory is also used, but out on
394 This file is only present if the CONFIG_MMU kernel configuration option is
397 The /proc/PID/clear_refs is used to reset the PG_Referenced and ACCESSED/YOUNG
398 bits on both physical and virtual pages associated with a process.
399 To clear the bits for all the pages associated with the process
400 > echo 1 > /proc/PID/clear_refs
402 To clear the bits for the anonymous pages associated with the process
403 > echo 2 > /proc/PID/clear_refs
405 To clear the bits for the file mapped pages associated with the process
406 > echo 3 > /proc/PID/clear_refs
407 Any other value written to /proc/PID/clear_refs will have no effect.
409 The /proc/pid/pagemap gives the PFN, which can be used to find the pageflags
410 using /proc/kpageflags and number of times a page is mapped using
411 /proc/kpagecount. For detailed explanation, see Documentation/vm/pagemap.txt.
416 Similar to the process entries, the kernel data files give information about
417 the running kernel. The files used to obtain this information are contained in
418 /proc and are listed in Table 1-5. Not all of these will be present in your
419 system. It depends on the kernel configuration and the loaded modules, which
420 files are there, and which are missing.
422 Table 1-5: Kernel info in /proc
423 ..............................................................................
425 apm Advanced power management info
426 buddyinfo Kernel memory allocator information (see text) (2.5)
427 bus Directory containing bus specific information
428 cmdline Kernel command line
429 cpuinfo Info about the CPU
430 devices Available devices (block and character)
431 dma Used DMS channels
432 filesystems Supported filesystems
433 driver Various drivers grouped here, currently rtc (2.4)
434 execdomains Execdomains, related to security (2.4)
435 fb Frame Buffer devices (2.4)
436 fs File system parameters, currently nfs/exports (2.4)
437 ide Directory containing info about the IDE subsystem
438 interrupts Interrupt usage
439 iomem Memory map (2.4)
440 ioports I/O port usage
441 irq Masks for irq to cpu affinity (2.4)(smp?)
442 isapnp ISA PnP (Plug&Play) Info (2.4)
443 kcore Kernel core image (can be ELF or A.OUT(deprecated in 2.4))
445 ksyms Kernel symbol table
446 loadavg Load average of last 1, 5 & 15 minutes
450 modules List of loaded modules
451 mounts Mounted filesystems
452 net Networking info (see text)
453 pagetypeinfo Additional page allocator information (see text) (2.5)
454 partitions Table of partitions known to the system
455 pci Deprecated info of PCI bus (new way -> /proc/bus/pci/,
456 decoupled by lspci (2.4)
458 scsi SCSI info (see text)
459 slabinfo Slab pool info
460 softirqs softirq usage
461 stat Overall statistics
462 swaps Swap space utilization
464 sysvipc Info of SysVIPC Resources (msg, sem, shm) (2.4)
465 tty Info of tty drivers
467 version Kernel version
468 video bttv info of video resources (2.4)
469 vmallocinfo Show vmalloced areas
470 ..............................................................................
472 You can, for example, check which interrupts are currently in use and what
473 they are used for by looking in the file /proc/interrupts:
475 > cat /proc/interrupts
477 0: 8728810 XT-PIC timer
478 1: 895 XT-PIC keyboard
480 3: 531695 XT-PIC aha152x
481 4: 2014133 XT-PIC serial
482 5: 44401 XT-PIC pcnet_cs
485 12: 182918 XT-PIC PS/2 Mouse
487 14: 1232265 XT-PIC ide0
491 In 2.4.* a couple of lines where added to this file LOC & ERR (this time is the
492 output of a SMP machine):
494 > cat /proc/interrupts
497 0: 1243498 1214548 IO-APIC-edge timer
498 1: 8949 8958 IO-APIC-edge keyboard
499 2: 0 0 XT-PIC cascade
500 5: 11286 10161 IO-APIC-edge soundblaster
501 8: 1 0 IO-APIC-edge rtc
502 9: 27422 27407 IO-APIC-edge 3c503
503 12: 113645 113873 IO-APIC-edge PS/2 Mouse
505 14: 22491 24012 IO-APIC-edge ide0
506 15: 2183 2415 IO-APIC-edge ide1
507 17: 30564 30414 IO-APIC-level eth0
508 18: 177 164 IO-APIC-level bttv
513 NMI is incremented in this case because every timer interrupt generates a NMI
514 (Non Maskable Interrupt) which is used by the NMI Watchdog to detect lockups.
516 LOC is the local interrupt counter of the internal APIC of every CPU.
518 ERR is incremented in the case of errors in the IO-APIC bus (the bus that
519 connects the CPUs in a SMP system. This means that an error has been detected,
520 the IO-APIC automatically retry the transmission, so it should not be a big
521 problem, but you should read the SMP-FAQ.
523 In 2.6.2* /proc/interrupts was expanded again. This time the goal was for
524 /proc/interrupts to display every IRQ vector in use by the system, not
525 just those considered 'most important'. The new vectors are:
527 THR -- interrupt raised when a machine check threshold counter
528 (typically counting ECC corrected errors of memory or cache) exceeds
529 a configurable threshold. Only available on some systems.
531 TRM -- a thermal event interrupt occurs when a temperature threshold
532 has been exceeded for the CPU. This interrupt may also be generated
533 when the temperature drops back to normal.
535 SPU -- a spurious interrupt is some interrupt that was raised then lowered
536 by some IO device before it could be fully processed by the APIC. Hence
537 the APIC sees the interrupt but does not know what device it came from.
538 For this case the APIC will generate the interrupt with a IRQ vector
539 of 0xff. This might also be generated by chipset bugs.
541 RES, CAL, TLB -- rescheduling, call and TLB flush interrupts are
542 sent from one CPU to another per the needs of the OS. Typically,
543 their statistics are used by kernel developers and interested users to
544 determine the occurrence of interrupts of the given type.
546 The above IRQ vectors are displayed only when relevant. For example,
547 the threshold vector does not exist on x86_64 platforms. Others are
548 suppressed when the system is a uniprocessor. As of this writing, only
549 i386 and x86_64 platforms support the new IRQ vector displays.
551 Of some interest is the introduction of the /proc/irq directory to 2.4.
552 It could be used to set IRQ to CPU affinity, this means that you can "hook" an
553 IRQ to only one CPU, or to exclude a CPU of handling IRQs. The contents of the
554 irq subdir is one subdir for each IRQ, and two files; default_smp_affinity and
559 0 10 12 14 16 18 2 4 6 8 prof_cpu_mask
560 1 11 13 15 17 19 3 5 7 9 default_smp_affinity
564 smp_affinity is a bitmask, in which you can specify which CPUs can handle the
565 IRQ, you can set it by doing:
567 > echo 1 > /proc/irq/10/smp_affinity
569 This means that only the first CPU will handle the IRQ, but you can also echo
570 5 which means that only the first and fourth CPU can handle the IRQ.
572 The contents of each smp_affinity file is the same by default:
574 > cat /proc/irq/0/smp_affinity
577 There is an alternate interface, smp_affinity_list which allows specifying
578 a cpu range instead of a bitmask:
580 > cat /proc/irq/0/smp_affinity_list
583 The default_smp_affinity mask applies to all non-active IRQs, which are the
584 IRQs which have not yet been allocated/activated, and hence which lack a
585 /proc/irq/[0-9]* directory.
587 The node file on an SMP system shows the node to which the device using the IRQ
588 reports itself as being attached. This hardware locality information does not
589 include information about any possible driver locality preference.
591 prof_cpu_mask specifies which CPUs are to be profiled by the system wide
592 profiler. Default value is ffffffff (all cpus if there are only 32 of them).
594 The way IRQs are routed is handled by the IO-APIC, and it's Round Robin
595 between all the CPUs which are allowed to handle it. As usual the kernel has
596 more info than you and does a better job than you, so the defaults are the
597 best choice for almost everyone. [Note this applies only to those IO-APIC's
598 that support "Round Robin" interrupt distribution.]
600 There are three more important subdirectories in /proc: net, scsi, and sys.
601 The general rule is that the contents, or even the existence of these
602 directories, depend on your kernel configuration. If SCSI is not enabled, the
603 directory scsi may not exist. The same is true with the net, which is there
604 only when networking support is present in the running kernel.
606 The slabinfo file gives information about memory usage at the slab level.
607 Linux uses slab pools for memory management above page level in version 2.2.
608 Commonly used objects have their own slab pool (such as network buffers,
609 directory cache, and so on).
611 ..............................................................................
613 > cat /proc/buddyinfo
615 Node 0, zone DMA 0 4 5 4 4 3 ...
616 Node 0, zone Normal 1 0 0 1 101 8 ...
617 Node 0, zone HighMem 2 0 0 1 1 0 ...
619 External fragmentation is a problem under some workloads, and buddyinfo is a
620 useful tool for helping diagnose these problems. Buddyinfo will give you a
621 clue as to how big an area you can safely allocate, or why a previous
624 Each column represents the number of pages of a certain order which are
625 available. In this case, there are 0 chunks of 2^0*PAGE_SIZE available in
626 ZONE_DMA, 4 chunks of 2^1*PAGE_SIZE in ZONE_DMA, 101 chunks of 2^4*PAGE_SIZE
627 available in ZONE_NORMAL, etc...
629 More information relevant to external fragmentation can be found in
632 > cat /proc/pagetypeinfo
636 Free pages count per migrate type at order 0 1 2 3 4 5 6 7 8 9 10
637 Node 0, zone DMA, type Unmovable 0 0 0 1 1 1 1 1 1 1 0
638 Node 0, zone DMA, type Reclaimable 0 0 0 0 0 0 0 0 0 0 0
639 Node 0, zone DMA, type Movable 1 1 2 1 2 1 1 0 1 0 2
640 Node 0, zone DMA, type Reserve 0 0 0 0 0 0 0 0 0 1 0
641 Node 0, zone DMA, type Isolate 0 0 0 0 0 0 0 0 0 0 0
642 Node 0, zone DMA32, type Unmovable 103 54 77 1 1 1 11 8 7 1 9
643 Node 0, zone DMA32, type Reclaimable 0 0 2 1 0 0 0 0 1 0 0
644 Node 0, zone DMA32, type Movable 169 152 113 91 77 54 39 13 6 1 452
645 Node 0, zone DMA32, type Reserve 1 2 2 2 2 0 1 1 1 1 0
646 Node 0, zone DMA32, type Isolate 0 0 0 0 0 0 0 0 0 0 0
648 Number of blocks type Unmovable Reclaimable Movable Reserve Isolate
649 Node 0, zone DMA 2 0 5 1 0
650 Node 0, zone DMA32 41 6 967 2 0
652 Fragmentation avoidance in the kernel works by grouping pages of different
653 migrate types into the same contiguous regions of memory called page blocks.
654 A page block is typically the size of the default hugepage size e.g. 2MB on
655 X86-64. By keeping pages grouped based on their ability to move, the kernel
656 can reclaim pages within a page block to satisfy a high-order allocation.
658 The pagetypinfo begins with information on the size of a page block. It
659 then gives the same type of information as buddyinfo except broken down
660 by migrate-type and finishes with details on how many page blocks of each
663 If min_free_kbytes has been tuned correctly (recommendations made by hugeadm
664 from libhugetlbfs http://sourceforge.net/projects/libhugetlbfs/), one can
665 make an estimate of the likely number of huge pages that can be allocated
666 at a given point in time. All the "Movable" blocks should be allocatable
667 unless memory has been mlock()'d. Some of the Reclaimable blocks should
668 also be allocatable although a lot of filesystem metadata may have to be
669 reclaimed to achieve this.
671 ..............................................................................
675 Provides information about distribution and utilization of memory. This
676 varies by architecture and compile options. The following is from a
677 16GB PIII, which has highmem enabled. You may not have all of these fields.
681 The "Locked" indicates whether the mapping is locked in memory or not.
684 MemTotal: 16344972 kB
691 HighTotal: 15597528 kB
692 HighFree: 13629632 kB
702 SReclaimable: 159856 kB
703 SUnreclaim: 124508 kB
708 CommitLimit: 7669796 kB
709 Committed_AS: 100056 kB
710 VmallocTotal: 112216 kB
712 VmallocChunk: 111088 kB
714 MemTotal: Total usable ram (i.e. physical ram minus a few reserved
715 bits and the kernel binary code)
716 MemFree: The sum of LowFree+HighFree
717 Buffers: Relatively temporary storage for raw disk blocks
718 shouldn't get tremendously large (20MB or so)
719 Cached: in-memory cache for files read from the disk (the
720 pagecache). Doesn't include SwapCached
721 SwapCached: Memory that once was swapped out, is swapped back in but
722 still also is in the swapfile (if memory is needed it
723 doesn't need to be swapped out AGAIN because it is already
724 in the swapfile. This saves I/O)
725 Active: Memory that has been used more recently and usually not
726 reclaimed unless absolutely necessary.
727 Inactive: Memory which has been less recently used. It is more
728 eligible to be reclaimed for other purposes
730 HighFree: Highmem is all memory above ~860MB of physical memory
731 Highmem areas are for use by userspace programs, or
732 for the pagecache. The kernel must use tricks to access
733 this memory, making it slower to access than lowmem.
735 LowFree: Lowmem is memory which can be used for everything that
736 highmem can be used for, but it is also available for the
737 kernel's use for its own data structures. Among many
738 other things, it is where everything from the Slab is
739 allocated. Bad things happen when you're out of lowmem.
740 SwapTotal: total amount of swap space available
741 SwapFree: Memory which has been evicted from RAM, and is temporarily
743 Dirty: Memory which is waiting to get written back to the disk
744 Writeback: Memory which is actively being written back to the disk
745 AnonPages: Non-file backed pages mapped into userspace page tables
746 Mapped: files which have been mmaped, such as libraries
747 Slab: in-kernel data structures cache
748 SReclaimable: Part of Slab, that might be reclaimed, such as caches
749 SUnreclaim: Part of Slab, that cannot be reclaimed on memory pressure
750 PageTables: amount of memory dedicated to the lowest level of page
752 NFS_Unstable: NFS pages sent to the server, but not yet committed to stable
754 Bounce: Memory used for block device "bounce buffers"
755 WritebackTmp: Memory used by FUSE for temporary writeback buffers
756 CommitLimit: Based on the overcommit ratio ('vm.overcommit_ratio'),
757 this is the total amount of memory currently available to
758 be allocated on the system. This limit is only adhered to
759 if strict overcommit accounting is enabled (mode 2 in
760 'vm.overcommit_memory').
761 The CommitLimit is calculated with the following formula:
762 CommitLimit = ('vm.overcommit_ratio' * Physical RAM) + Swap
763 For example, on a system with 1G of physical RAM and 7G
764 of swap with a `vm.overcommit_ratio` of 30 it would
765 yield a CommitLimit of 7.3G.
766 For more details, see the memory overcommit documentation
767 in vm/overcommit-accounting.
768 Committed_AS: The amount of memory presently allocated on the system.
769 The committed memory is a sum of all of the memory which
770 has been allocated by processes, even if it has not been
771 "used" by them as of yet. A process which malloc()'s 1G
772 of memory, but only touches 300M of it will only show up
773 as using 300M of memory even if it has the address space
774 allocated for the entire 1G. This 1G is memory which has
775 been "committed" to by the VM and can be used at any time
776 by the allocating application. With strict overcommit
777 enabled on the system (mode 2 in 'vm.overcommit_memory'),
778 allocations which would exceed the CommitLimit (detailed
779 above) will not be permitted. This is useful if one needs
780 to guarantee that processes will not fail due to lack of
781 memory once that memory has been successfully allocated.
782 VmallocTotal: total size of vmalloc memory area
783 VmallocUsed: amount of vmalloc area which is used
784 VmallocChunk: largest contiguous block of vmalloc area which is free
786 ..............................................................................
790 Provides information about vmalloced/vmaped areas. One line per area,
791 containing the virtual address range of the area, size in bytes,
792 caller information of the creator, and optional information depending
793 on the kind of area :
795 pages=nr number of pages
796 phys=addr if a physical address was specified
797 ioremap I/O mapping (ioremap() and friends)
798 vmalloc vmalloc() area
801 vpages buffer for pages pointers was vmalloced (huge area)
802 N<node>=nr (Only on NUMA kernels)
803 Number of pages allocated on memory node <node>
805 > cat /proc/vmallocinfo
806 0xffffc20000000000-0xffffc20000201000 2101248 alloc_large_system_hash+0x204 ...
807 /0x2c0 pages=512 vmalloc N0=128 N1=128 N2=128 N3=128
808 0xffffc20000201000-0xffffc20000302000 1052672 alloc_large_system_hash+0x204 ...
809 /0x2c0 pages=256 vmalloc N0=64 N1=64 N2=64 N3=64
810 0xffffc20000302000-0xffffc20000304000 8192 acpi_tb_verify_table+0x21/0x4f...
811 phys=7fee8000 ioremap
812 0xffffc20000304000-0xffffc20000307000 12288 acpi_tb_verify_table+0x21/0x4f...
813 phys=7fee7000 ioremap
814 0xffffc2000031d000-0xffffc2000031f000 8192 init_vdso_vars+0x112/0x210
815 0xffffc2000031f000-0xffffc2000032b000 49152 cramfs_uncompress_init+0x2e ...
816 /0x80 pages=11 vmalloc N0=3 N1=3 N2=2 N3=3
817 0xffffc2000033a000-0xffffc2000033d000 12288 sys_swapon+0x640/0xac0 ...
819 0xffffc20000347000-0xffffc2000034c000 20480 xt_alloc_table_info+0xfe ...
820 /0x130 [x_tables] pages=4 vmalloc N0=4
821 0xffffffffa0000000-0xffffffffa000f000 61440 sys_init_module+0xc27/0x1d00 ...
822 pages=14 vmalloc N2=14
823 0xffffffffa000f000-0xffffffffa0014000 20480 sys_init_module+0xc27/0x1d00 ...
825 0xffffffffa0014000-0xffffffffa0017000 12288 sys_init_module+0xc27/0x1d00 ...
827 0xffffffffa0017000-0xffffffffa0022000 45056 sys_init_module+0xc27/0x1d00 ...
828 pages=10 vmalloc N0=10
830 ..............................................................................
834 Provides counts of softirq handlers serviced since boot time, for each cpu.
839 TIMER: 27166 27120 27097 27034
844 SCHED: 27035 26983 26971 26746
848 1.3 IDE devices in /proc/ide
849 ----------------------------
851 The subdirectory /proc/ide contains information about all IDE devices of which
852 the kernel is aware. There is one subdirectory for each IDE controller, the
853 file drivers and a link for each IDE device, pointing to the device directory
854 in the controller specific subtree.
856 The file drivers contains general information about the drivers used for the
859 > cat /proc/ide/drivers
860 ide-cdrom version 4.53
861 ide-disk version 1.08
863 More detailed information can be found in the controller specific
864 subdirectories. These are named ide0, ide1 and so on. Each of these
865 directories contains the files shown in table 1-6.
868 Table 1-6: IDE controller info in /proc/ide/ide?
869 ..............................................................................
871 channel IDE channel (0 or 1)
872 config Configuration (only for PCI/IDE bridge)
874 model Type/Chipset of IDE controller
875 ..............................................................................
877 Each device connected to a controller has a separate subdirectory in the
878 controllers directory. The files listed in table 1-7 are contained in these
882 Table 1-7: IDE device information
883 ..............................................................................
886 capacity Capacity of the medium (in 512Byte blocks)
887 driver driver and version
888 geometry physical and logical geometry
889 identify device identify block
891 model device identifier
892 settings device setup
893 smart_thresholds IDE disk management thresholds
894 smart_values IDE disk management values
895 ..............................................................................
897 The most interesting file is settings. This file contains a nice overview of
898 the drive parameters:
900 # cat /proc/ide/ide0/hda/settings
901 name value min max mode
902 ---- ----- --- --- ----
903 bios_cyl 526 0 65535 rw
904 bios_head 255 0 255 rw
906 breada_readahead 4 0 127 rw
908 file_readahead 72 0 2097151 rw
910 keepsettings 0 0 1 rw
911 max_kb_per_request 122 1 127 rw
915 pio_mode write-only 0 255 w
921 1.4 Networking info in /proc/net
922 --------------------------------
924 The subdirectory /proc/net follows the usual pattern. Table 1-8 shows the
925 additional values you get for IP version 6 if you configure the kernel to
926 support this. Table 1-9 lists the files and their meaning.
929 Table 1-8: IPv6 info in /proc/net
930 ..............................................................................
932 udp6 UDP sockets (IPv6)
933 tcp6 TCP sockets (IPv6)
934 raw6 Raw device statistics (IPv6)
935 igmp6 IP multicast addresses, which this host joined (IPv6)
936 if_inet6 List of IPv6 interface addresses
937 ipv6_route Kernel routing table for IPv6
938 rt6_stats Global IPv6 routing tables statistics
939 sockstat6 Socket statistics (IPv6)
940 snmp6 Snmp data (IPv6)
941 ..............................................................................
944 Table 1-9: Network info in /proc/net
945 ..............................................................................
948 dev network devices with statistics
949 dev_mcast the Layer2 multicast groups a device is listening too
950 (interface index, label, number of references, number of bound
952 dev_stat network device status
953 ip_fwchains Firewall chain linkage
954 ip_fwnames Firewall chain names
955 ip_masq Directory containing the masquerading tables
956 ip_masquerade Major masquerading table
957 netstat Network statistics
958 raw raw device statistics
959 route Kernel routing table
960 rpc Directory containing rpc info
961 rt_cache Routing cache
963 sockstat Socket statistics
965 tr_rif Token ring RIF routing table
967 unix UNIX domain sockets
968 wireless Wireless interface data (Wavelan etc)
969 igmp IP multicast addresses, which this host joined
970 psched Global packet scheduler parameters.
971 netlink List of PF_NETLINK sockets
972 ip_mr_vifs List of multicast virtual interfaces
973 ip_mr_cache List of multicast routing cache
974 ..............................................................................
976 You can use this information to see which network devices are available in
977 your system and how much traffic was routed over those devices:
981 face |bytes packets errs drop fifo frame compressed multicast|[...
982 lo: 908188 5596 0 0 0 0 0 0 [...
983 ppp0:15475140 20721 410 0 0 410 0 0 [...
984 eth0: 614530 7085 0 0 0 0 0 1 [...
987 ...] bytes packets errs drop fifo colls carrier compressed
988 ...] 908188 5596 0 0 0 0 0 0
989 ...] 1375103 17405 0 0 0 0 0 0
990 ...] 1703981 5535 0 0 0 3 0 0
992 In addition, each Channel Bond interface has its own directory. For
993 example, the bond0 device will have a directory called /proc/net/bond0/.
994 It will contain information that is specific to that bond, such as the
995 current slaves of the bond, the link status of the slaves, and how
996 many times the slaves link has failed.
1001 If you have a SCSI host adapter in your system, you'll find a subdirectory
1002 named after the driver for this adapter in /proc/scsi. You'll also see a list
1003 of all recognized SCSI devices in /proc/scsi:
1005 >cat /proc/scsi/scsi
1007 Host: scsi0 Channel: 00 Id: 00 Lun: 00
1008 Vendor: IBM Model: DGHS09U Rev: 03E0
1009 Type: Direct-Access ANSI SCSI revision: 03
1010 Host: scsi0 Channel: 00 Id: 06 Lun: 00
1011 Vendor: PIONEER Model: CD-ROM DR-U06S Rev: 1.04
1012 Type: CD-ROM ANSI SCSI revision: 02
1015 The directory named after the driver has one file for each adapter found in
1016 the system. These files contain information about the controller, including
1017 the used IRQ and the IO address range. The amount of information shown is
1018 dependent on the adapter you use. The example shows the output for an Adaptec
1019 AHA-2940 SCSI adapter:
1021 > cat /proc/scsi/aic7xxx/0
1023 Adaptec AIC7xxx driver version: 5.1.19/3.2.4
1025 TCQ Enabled By Default : Disabled
1026 AIC7XXX_PROC_STATS : Disabled
1027 AIC7XXX_RESET_DELAY : 5
1028 Adapter Configuration:
1029 SCSI Adapter: Adaptec AHA-294X Ultra SCSI host adapter
1030 Ultra Wide Controller
1031 PCI MMAPed I/O Base: 0xeb001000
1032 Adapter SEEPROM Config: SEEPROM found and used.
1033 Adaptec SCSI BIOS: Enabled
1035 SCBs: Active 0, Max Active 2,
1036 Allocated 15, HW 16, Page 255
1038 BIOS Control Word: 0x18b6
1039 Adapter Control Word: 0x005b
1040 Extended Translation: Enabled
1041 Disconnect Enable Flags: 0xffff
1042 Ultra Enable Flags: 0x0001
1043 Tag Queue Enable Flags: 0x0000
1044 Ordered Queue Tag Flags: 0x0000
1045 Default Tag Queue Depth: 8
1046 Tagged Queue By Device array for aic7xxx host instance 0:
1047 {255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255}
1048 Actual queue depth per device for aic7xxx host instance 0:
1049 {1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1}
1052 Device using Wide/Sync transfers at 40.0 MByte/sec, offset 8
1053 Transinfo settings: current(12/8/1/0), goal(12/8/1/0), user(12/15/1/0)
1054 Total transfers 160151 (74577 reads and 85574 writes)
1056 Device using Narrow/Sync transfers at 5.0 MByte/sec, offset 15
1057 Transinfo settings: current(50/15/0/0), goal(50/15/0/0), user(50/15/0/0)
1058 Total transfers 0 (0 reads and 0 writes)
1061 1.6 Parallel port info in /proc/parport
1062 ---------------------------------------
1064 The directory /proc/parport contains information about the parallel ports of
1065 your system. It has one subdirectory for each port, named after the port
1068 These directories contain the four files shown in Table 1-10.
1071 Table 1-10: Files in /proc/parport
1072 ..............................................................................
1074 autoprobe Any IEEE-1284 device ID information that has been acquired.
1075 devices list of the device drivers using that port. A + will appear by the
1076 name of the device currently using the port (it might not appear
1078 hardware Parallel port's base address, IRQ line and DMA channel.
1079 irq IRQ that parport is using for that port. This is in a separate
1080 file to allow you to alter it by writing a new value in (IRQ
1082 ..............................................................................
1084 1.7 TTY info in /proc/tty
1085 -------------------------
1087 Information about the available and actually used tty's can be found in the
1088 directory /proc/tty.You'll find entries for drivers and line disciplines in
1089 this directory, as shown in Table 1-11.
1092 Table 1-11: Files in /proc/tty
1093 ..............................................................................
1095 drivers list of drivers and their usage
1096 ldiscs registered line disciplines
1097 driver/serial usage statistic and status of single tty lines
1098 ..............................................................................
1100 To see which tty's are currently in use, you can simply look into the file
1103 > cat /proc/tty/drivers
1104 pty_slave /dev/pts 136 0-255 pty:slave
1105 pty_master /dev/ptm 128 0-255 pty:master
1106 pty_slave /dev/ttyp 3 0-255 pty:slave
1107 pty_master /dev/pty 2 0-255 pty:master
1108 serial /dev/cua 5 64-67 serial:callout
1109 serial /dev/ttyS 4 64-67 serial
1110 /dev/tty0 /dev/tty0 4 0 system:vtmaster
1111 /dev/ptmx /dev/ptmx 5 2 system
1112 /dev/console /dev/console 5 1 system:console
1113 /dev/tty /dev/tty 5 0 system:/dev/tty
1114 unknown /dev/tty 4 1-63 console
1117 1.8 Miscellaneous kernel statistics in /proc/stat
1118 -------------------------------------------------
1120 Various pieces of information about kernel activity are available in the
1121 /proc/stat file. All of the numbers reported in this file are aggregates
1122 since the system first booted. For a quick look, simply cat the file:
1125 cpu 2255 34 2290 22625563 6290 127 456 0 0
1126 cpu0 1132 34 1441 11311718 3675 127 438 0 0
1127 cpu1 1123 0 849 11313845 2614 0 18 0 0
1128 intr 114930548 113199788 3 0 5 263 0 4 [... lots more numbers ...]
1134 softirq 183433 0 21755 12 39 1137 231 21459 2263
1136 The very first "cpu" line aggregates the numbers in all of the other "cpuN"
1137 lines. These numbers identify the amount of time the CPU has spent performing
1138 different kinds of work. Time units are in USER_HZ (typically hundredths of a
1139 second). The meanings of the columns are as follows, from left to right:
1141 - user: normal processes executing in user mode
1142 - nice: niced processes executing in user mode
1143 - system: processes executing in kernel mode
1144 - idle: twiddling thumbs
1145 - iowait: waiting for I/O to complete
1146 - irq: servicing interrupts
1147 - softirq: servicing softirqs
1148 - steal: involuntary wait
1149 - guest: running a normal guest
1150 - guest_nice: running a niced guest
1152 The "intr" line gives counts of interrupts serviced since boot time, for each
1153 of the possible system interrupts. The first column is the total of all
1154 interrupts serviced; each subsequent column is the total for that particular
1157 The "ctxt" line gives the total number of context switches across all CPUs.
1159 The "btime" line gives the time at which the system booted, in seconds since
1162 The "processes" line gives the number of processes and threads created, which
1163 includes (but is not limited to) those created by calls to the fork() and
1164 clone() system calls.
1166 The "procs_running" line gives the total number of threads that are
1167 running or ready to run (i.e., the total number of runnable threads).
1169 The "procs_blocked" line gives the number of processes currently blocked,
1170 waiting for I/O to complete.
1172 The "softirq" line gives counts of softirqs serviced since boot time, for each
1173 of the possible system softirqs. The first column is the total of all
1174 softirqs serviced; each subsequent column is the total for that particular
1178 1.9 Ext4 file system parameters
1179 ------------------------------
1181 Information about mounted ext4 file systems can be found in
1182 /proc/fs/ext4. Each mounted filesystem will have a directory in
1183 /proc/fs/ext4 based on its device name (i.e., /proc/fs/ext4/hdc or
1184 /proc/fs/ext4/dm-0). The files in each per-device directory are shown
1185 in Table 1-12, below.
1187 Table 1-12: Files in /proc/fs/ext4/<devname>
1188 ..............................................................................
1190 mb_groups details of multiblock allocator buddy cache of free blocks
1191 ..............................................................................
1195 Shows registered system console lines.
1197 To see which character device lines are currently used for the system console
1198 /dev/console, you may simply look into the file /proc/consoles:
1200 > cat /proc/consoles
1206 device name of the device
1207 operations R = can do read operations
1208 W = can do write operations
1210 flags E = it is enabled
1211 C = it is preferred console
1212 B = it is primary boot console
1213 p = it is used for printk buffer
1214 b = it is not a TTY but a Braille device
1215 a = it is safe to use when cpu is offline
1216 major:minor major and minor number of the device separated by a colon
1218 ------------------------------------------------------------------------------
1220 ------------------------------------------------------------------------------
1221 The /proc file system serves information about the running system. It not only
1222 allows access to process data but also allows you to request the kernel status
1223 by reading files in the hierarchy.
1225 The directory structure of /proc reflects the types of information and makes
1226 it easy, if not obvious, where to look for specific data.
1227 ------------------------------------------------------------------------------
1229 ------------------------------------------------------------------------------
1230 CHAPTER 2: MODIFYING SYSTEM PARAMETERS
1231 ------------------------------------------------------------------------------
1233 ------------------------------------------------------------------------------
1235 ------------------------------------------------------------------------------
1236 * Modifying kernel parameters by writing into files found in /proc/sys
1237 * Exploring the files which modify certain parameters
1238 * Review of the /proc/sys file tree
1239 ------------------------------------------------------------------------------
1242 A very interesting part of /proc is the directory /proc/sys. This is not only
1243 a source of information, it also allows you to change parameters within the
1244 kernel. Be very careful when attempting this. You can optimize your system,
1245 but you can also cause it to crash. Never alter kernel parameters on a
1246 production system. Set up a development machine and test to make sure that
1247 everything works the way you want it to. You may have no alternative but to
1248 reboot the machine once an error has been made.
1250 To change a value, simply echo the new value into the file. An example is
1251 given below in the section on the file system data. You need to be root to do
1252 this. You can create your own boot script to perform this every time your
1255 The files in /proc/sys can be used to fine tune and monitor miscellaneous and
1256 general things in the operation of the Linux kernel. Since some of the files
1257 can inadvertently disrupt your system, it is advisable to read both
1258 documentation and source before actually making adjustments. In any case, be
1259 very careful when writing to any of these files. The entries in /proc may
1260 change slightly between the 2.1.* and the 2.2 kernel, so if there is any doubt
1261 review the kernel documentation in the directory /usr/src/linux/Documentation.
1262 This chapter is heavily based on the documentation included in the pre 2.2
1263 kernels, and became part of it in version 2.2.1 of the Linux kernel.
1265 Please see: Documentation/sysctls/ directory for descriptions of these
1268 ------------------------------------------------------------------------------
1270 ------------------------------------------------------------------------------
1271 Certain aspects of kernel behavior can be modified at runtime, without the
1272 need to recompile the kernel, or even to reboot the system. The files in the
1273 /proc/sys tree can not only be read, but also modified. You can use the echo
1274 command to write value into these files, thereby changing the default settings
1276 ------------------------------------------------------------------------------
1278 ------------------------------------------------------------------------------
1279 CHAPTER 3: PER-PROCESS PARAMETERS
1280 ------------------------------------------------------------------------------
1282 3.1 /proc/<pid>/oom_adj & /proc/<pid>/oom_score_adj- Adjust the oom-killer score
1283 --------------------------------------------------------------------------------
1285 These file can be used to adjust the badness heuristic used to select which
1286 process gets killed in out of memory conditions.
1288 The badness heuristic assigns a value to each candidate task ranging from 0
1289 (never kill) to 1000 (always kill) to determine which process is targeted. The
1290 units are roughly a proportion along that range of allowed memory the process
1291 may allocate from based on an estimation of its current memory and swap use.
1292 For example, if a task is using all allowed memory, its badness score will be
1293 1000. If it is using half of its allowed memory, its score will be 500.
1295 There is an additional factor included in the badness score: root
1296 processes are given 3% extra memory over other tasks.
1298 The amount of "allowed" memory depends on the context in which the oom killer
1299 was called. If it is due to the memory assigned to the allocating task's cpuset
1300 being exhausted, the allowed memory represents the set of mems assigned to that
1301 cpuset. If it is due to a mempolicy's node(s) being exhausted, the allowed
1302 memory represents the set of mempolicy nodes. If it is due to a memory
1303 limit (or swap limit) being reached, the allowed memory is that configured
1304 limit. Finally, if it is due to the entire system being out of memory, the
1305 allowed memory represents all allocatable resources.
1307 The value of /proc/<pid>/oom_score_adj is added to the badness score before it
1308 is used to determine which task to kill. Acceptable values range from -1000
1309 (OOM_SCORE_ADJ_MIN) to +1000 (OOM_SCORE_ADJ_MAX). This allows userspace to
1310 polarize the preference for oom killing either by always preferring a certain
1311 task or completely disabling it. The lowest possible value, -1000, is
1312 equivalent to disabling oom killing entirely for that task since it will always
1313 report a badness score of 0.
1315 Consequently, it is very simple for userspace to define the amount of memory to
1316 consider for each task. Setting a /proc/<pid>/oom_score_adj value of +500, for
1317 example, is roughly equivalent to allowing the remainder of tasks sharing the
1318 same system, cpuset, mempolicy, or memory controller resources to use at least
1319 50% more memory. A value of -500, on the other hand, would be roughly
1320 equivalent to discounting 50% of the task's allowed memory from being considered
1321 as scoring against the task.
1323 For backwards compatibility with previous kernels, /proc/<pid>/oom_adj may also
1324 be used to tune the badness score. Its acceptable values range from -16
1325 (OOM_ADJUST_MIN) to +15 (OOM_ADJUST_MAX) and a special value of -17
1326 (OOM_DISABLE) to disable oom killing entirely for that task. Its value is
1327 scaled linearly with /proc/<pid>/oom_score_adj.
1329 Writing to /proc/<pid>/oom_score_adj or /proc/<pid>/oom_adj will change the
1330 other with its scaled value.
1332 The value of /proc/<pid>/oom_score_adj may be reduced no lower than the last
1333 value set by a CAP_SYS_RESOURCE process. To reduce the value any lower
1334 requires CAP_SYS_RESOURCE.
1336 NOTICE: /proc/<pid>/oom_adj is deprecated and will be removed, please see
1337 Documentation/feature-removal-schedule.txt.
1339 Caveat: when a parent task is selected, the oom killer will sacrifice any first
1340 generation children with separate address spaces instead, if possible. This
1341 avoids servers and important system daemons from being killed and loses the
1342 minimal amount of work.
1345 3.2 /proc/<pid>/oom_score - Display current oom-killer score
1346 -------------------------------------------------------------
1348 This file can be used to check the current score used by the oom-killer is for
1349 any given <pid>. Use it together with /proc/<pid>/oom_adj to tune which
1350 process should be killed in an out-of-memory situation.
1353 3.3 /proc/<pid>/io - Display the IO accounting fields
1354 -------------------------------------------------------
1356 This file contains IO statistics for each running process
1361 test:/tmp # dd if=/dev/zero of=/tmp/test.dat &
1364 test:/tmp # cat /proc/3828/io
1370 write_bytes: 323932160
1371 cancelled_write_bytes: 0
1380 I/O counter: chars read
1381 The number of bytes which this task has caused to be read from storage. This
1382 is simply the sum of bytes which this process passed to read() and pread().
1383 It includes things like tty IO and it is unaffected by whether or not actual
1384 physical disk IO was required (the read might have been satisfied from
1391 I/O counter: chars written
1392 The number of bytes which this task has caused, or shall cause to be written
1393 to disk. Similar caveats apply here as with rchar.
1399 I/O counter: read syscalls
1400 Attempt to count the number of read I/O operations, i.e. syscalls like read()
1407 I/O counter: write syscalls
1408 Attempt to count the number of write I/O operations, i.e. syscalls like
1409 write() and pwrite().
1415 I/O counter: bytes read
1416 Attempt to count the number of bytes which this process really did cause to
1417 be fetched from the storage layer. Done at the submit_bio() level, so it is
1418 accurate for block-backed filesystems. <please add status regarding NFS and
1419 CIFS at a later time>
1425 I/O counter: bytes written
1426 Attempt to count the number of bytes which this process caused to be sent to
1427 the storage layer. This is done at page-dirtying time.
1430 cancelled_write_bytes
1431 ---------------------
1433 The big inaccuracy here is truncate. If a process writes 1MB to a file and
1434 then deletes the file, it will in fact perform no writeout. But it will have
1435 been accounted as having caused 1MB of write.
1436 In other words: The number of bytes which this process caused to not happen,
1437 by truncating pagecache. A task can cause "negative" IO too. If this task
1438 truncates some dirty pagecache, some IO which another task has been accounted
1439 for (in its write_bytes) will not be happening. We _could_ just subtract that
1440 from the truncating task's write_bytes, but there is information loss in doing
1447 At its current implementation state, this is a bit racy on 32-bit machines: if
1448 process A reads process B's /proc/pid/io while process B is updating one of
1449 those 64-bit counters, process A could see an intermediate result.
1452 More information about this can be found within the taskstats documentation in
1453 Documentation/accounting.
1455 3.4 /proc/<pid>/coredump_filter - Core dump filtering settings
1456 ---------------------------------------------------------------
1457 When a process is dumped, all anonymous memory is written to a core file as
1458 long as the size of the core file isn't limited. But sometimes we don't want
1459 to dump some memory segments, for example, huge shared memory. Conversely,
1460 sometimes we want to save file-backed memory segments into a core file, not
1461 only the individual files.
1463 /proc/<pid>/coredump_filter allows you to customize which memory segments
1464 will be dumped when the <pid> process is dumped. coredump_filter is a bitmask
1465 of memory types. If a bit of the bitmask is set, memory segments of the
1466 corresponding memory type are dumped, otherwise they are not dumped.
1468 The following 7 memory types are supported:
1469 - (bit 0) anonymous private memory
1470 - (bit 1) anonymous shared memory
1471 - (bit 2) file-backed private memory
1472 - (bit 3) file-backed shared memory
1473 - (bit 4) ELF header pages in file-backed private memory areas (it is
1474 effective only if the bit 2 is cleared)
1475 - (bit 5) hugetlb private memory
1476 - (bit 6) hugetlb shared memory
1478 Note that MMIO pages such as frame buffer are never dumped and vDSO pages
1479 are always dumped regardless of the bitmask status.
1481 Note bit 0-4 doesn't effect any hugetlb memory. hugetlb memory are only
1482 effected by bit 5-6.
1484 Default value of coredump_filter is 0x23; this means all anonymous memory
1485 segments and hugetlb private memory are dumped.
1487 If you don't want to dump all shared memory segments attached to pid 1234,
1488 write 0x21 to the process's proc file.
1490 $ echo 0x21 > /proc/1234/coredump_filter
1492 When a new process is created, the process inherits the bitmask status from its
1493 parent. It is useful to set up coredump_filter before the program runs.
1496 $ echo 0x7 > /proc/self/coredump_filter
1499 3.5 /proc/<pid>/mountinfo - Information about mounts
1500 --------------------------------------------------------
1502 This file contains lines of the form:
1504 36 35 98:0 /mnt1 /mnt2 rw,noatime master:1 - ext3 /dev/root rw,errors=continue
1505 (1)(2)(3) (4) (5) (6) (7) (8) (9) (10) (11)
1507 (1) mount ID: unique identifier of the mount (may be reused after umount)
1508 (2) parent ID: ID of parent (or of self for the top of the mount tree)
1509 (3) major:minor: value of st_dev for files on filesystem
1510 (4) root: root of the mount within the filesystem
1511 (5) mount point: mount point relative to the process's root
1512 (6) mount options: per mount options
1513 (7) optional fields: zero or more fields of the form "tag[:value]"
1514 (8) separator: marks the end of the optional fields
1515 (9) filesystem type: name of filesystem of the form "type[.subtype]"
1516 (10) mount source: filesystem specific information or "none"
1517 (11) super options: per super block options
1519 Parsers should ignore all unrecognised optional fields. Currently the
1520 possible optional fields are:
1522 shared:X mount is shared in peer group X
1523 master:X mount is slave to peer group X
1524 propagate_from:X mount is slave and receives propagation from peer group X (*)
1525 unbindable mount is unbindable
1527 (*) X is the closest dominant peer group under the process's root. If
1528 X is the immediate master of the mount, or if there's no dominant peer
1529 group under the same root, then only the "master:X" field is present
1530 and not the "propagate_from:X" field.
1532 For more information on mount propagation see:
1534 Documentation/filesystems/sharedsubtree.txt
1537 3.6 /proc/<pid>/comm & /proc/<pid>/task/<tid>/comm
1538 --------------------------------------------------------
1539 These files provide a method to access a tasks comm value. It also allows for
1540 a task to set its own or one of its thread siblings comm value. The comm value
1541 is limited in size compared to the cmdline value, so writing anything longer
1542 then the kernel's TASK_COMM_LEN (currently 16 chars) will result in a truncated
Code Review / linux-2.6.git / blob