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 ------------------------------------------------------------------------------
9 Version 1.3 Kernel version 2.2.12
10 Kernel version 2.4.0-test11-pre4
11 ------------------------------------------------------------------------------
17 0.1 Introduction/Credits
20 1 Collecting System Information
21 1.1 Process-Specific Subdirectories
23 1.3 IDE devices in /proc/ide
24 1.4 Networking info in /proc/net
26 1.6 Parallel port info in /proc/parport
27 1.7 TTY info in /proc/tty
28 1.8 Miscellaneous kernel statistics in /proc/stat
30 2 Modifying System Parameters
31 2.1 /proc/sys/fs - File system data
32 2.2 /proc/sys/fs/binfmt_misc - Miscellaneous binary formats
33 2.3 /proc/sys/kernel - general kernel parameters
34 2.4 /proc/sys/vm - The virtual memory subsystem
35 2.5 /proc/sys/dev - Device specific parameters
36 2.6 /proc/sys/sunrpc - Remote procedure calls
37 2.7 /proc/sys/net - Networking stuff
38 2.8 /proc/sys/net/ipv4 - IPV4 settings
41 2.11 /proc/sys/fs/mqueue - POSIX message queues filesystem
42 2.12 /proc/<pid>/oom_adj - Adjust the oom-killer score
43 2.13 /proc/<pid>/oom_score - Display current oom-killer score
44 2.14 /proc/<pid>/io - Display the IO accounting fields
45 2.15 /proc/<pid>/coredump_filter - Core dump filtering settings
46 2.16 /proc/<pid>/mountinfo - Information about mounts
47 2.17 /proc/sys/fs/epoll - Configuration options for the epoll interface
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://skaro.nightcrawler.com/~bb/Docs/Proc as HTML version.
83 If the above direction does not works for you, ypu 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
143 smaps Extension based on maps, the rss size for each mapped file
144 ..............................................................................
146 For example, to get the status information of a process, all you have to do is
147 read the file /proc/PID/status:
149 >cat /proc/self/status
165 SigPnd: 0000000000000000
166 SigBlk: 0000000000000000
167 SigIgn: 0000000000000000
168 SigCgt: 0000000000000000
169 CapInh: 00000000fffffeff
170 CapPrm: 0000000000000000
171 CapEff: 0000000000000000
174 This shows you nearly the same information you would get if you viewed it with
175 the ps command. In fact, ps uses the proc file system to obtain its
176 information. The statm file contains more detailed information about the
177 process memory usage. Its seven fields are explained in Table 1-2. The stat
178 file contains details information about the process itself. Its fields are
179 explained in Table 1-3.
182 Table 1-2: Contents of the statm files (as of 2.6.8-rc3)
183 ..............................................................................
185 size total program size (pages) (same as VmSize in status)
186 resident size of memory portions (pages) (same as VmRSS in status)
187 shared number of pages that are shared (i.e. backed by a file)
188 trs number of pages that are 'code' (not including libs; broken,
189 includes data segment)
190 lrs number of pages of library (always 0 on 2.6)
191 drs number of pages of data/stack (including libs; broken,
192 includes library text)
193 dt number of dirty pages (always 0 on 2.6)
194 ..............................................................................
197 Table 1-3: Contents of the stat files (as of 2.6.22-rc3)
198 ..............................................................................
201 tcomm filename of the executable
202 state state (R is running, S is sleeping, D is sleeping in an
203 uninterruptible wait, Z is zombie, T is traced or stopped)
204 ppid process id of the parent process
205 pgrp pgrp of the process
207 tty_nr tty the process uses
208 tty_pgrp pgrp of the tty
210 min_flt number of minor faults
211 cmin_flt number of minor faults with child's
212 maj_flt number of major faults
213 cmaj_flt number of major faults with child's
214 utime user mode jiffies
215 stime kernel mode jiffies
216 cutime user mode jiffies with child's
217 cstime kernel mode jiffies with child's
218 priority priority level
220 num_threads number of threads
221 it_real_value (obsolete, always 0)
222 start_time time the process started after system boot
223 vsize virtual memory size
224 rss resident set memory size
225 rsslim current limit in bytes on the rss
226 start_code address above which program text can run
227 end_code address below which program text can run
228 start_stack address of the start of the stack
229 esp current value of ESP
230 eip current value of EIP
231 pending bitmap of pending signals (obsolete)
232 blocked bitmap of blocked signals (obsolete)
233 sigign bitmap of ignored signals (obsolete)
234 sigcatch bitmap of catched signals (obsolete)
235 wchan address where process went to sleep
238 exit_signal signal to send to parent thread on exit
239 task_cpu which CPU the task is scheduled on
240 rt_priority realtime priority
241 policy scheduling policy (man sched_setscheduler)
242 blkio_ticks time spent waiting for block IO
243 ..............................................................................
249 Similar to the process entries, the kernel data files give information about
250 the running kernel. The files used to obtain this information are contained in
251 /proc and are listed in Table 1-4. Not all of these will be present in your
252 system. It depends on the kernel configuration and the loaded modules, which
253 files are there, and which are missing.
255 Table 1-4: Kernel info in /proc
256 ..............................................................................
258 apm Advanced power management info
259 buddyinfo Kernel memory allocator information (see text) (2.5)
260 bus Directory containing bus specific information
261 cmdline Kernel command line
262 cpuinfo Info about the CPU
263 devices Available devices (block and character)
264 dma Used DMS channels
265 filesystems Supported filesystems
266 driver Various drivers grouped here, currently rtc (2.4)
267 execdomains Execdomains, related to security (2.4)
268 fb Frame Buffer devices (2.4)
269 fs File system parameters, currently nfs/exports (2.4)
270 ide Directory containing info about the IDE subsystem
271 interrupts Interrupt usage
272 iomem Memory map (2.4)
273 ioports I/O port usage
274 irq Masks for irq to cpu affinity (2.4)(smp?)
275 isapnp ISA PnP (Plug&Play) Info (2.4)
276 kcore Kernel core image (can be ELF or A.OUT(deprecated in 2.4))
278 ksyms Kernel symbol table
279 loadavg Load average of last 1, 5 & 15 minutes
283 modules List of loaded modules
284 mounts Mounted filesystems
285 net Networking info (see text)
286 partitions Table of partitions known to the system
287 pci Deprecated info of PCI bus (new way -> /proc/bus/pci/,
288 decoupled by lspci (2.4)
290 scsi SCSI info (see text)
291 slabinfo Slab pool info
292 stat Overall statistics
293 swaps Swap space utilization
295 sysvipc Info of SysVIPC Resources (msg, sem, shm) (2.4)
296 tty Info of tty drivers
298 version Kernel version
299 video bttv info of video resources (2.4)
300 vmallocinfo Show vmalloced areas
301 ..............................................................................
303 You can, for example, check which interrupts are currently in use and what
304 they are used for by looking in the file /proc/interrupts:
306 > cat /proc/interrupts
308 0: 8728810 XT-PIC timer
309 1: 895 XT-PIC keyboard
311 3: 531695 XT-PIC aha152x
312 4: 2014133 XT-PIC serial
313 5: 44401 XT-PIC pcnet_cs
316 12: 182918 XT-PIC PS/2 Mouse
318 14: 1232265 XT-PIC ide0
322 In 2.4.* a couple of lines where added to this file LOC & ERR (this time is the
323 output of a SMP machine):
325 > cat /proc/interrupts
328 0: 1243498 1214548 IO-APIC-edge timer
329 1: 8949 8958 IO-APIC-edge keyboard
330 2: 0 0 XT-PIC cascade
331 5: 11286 10161 IO-APIC-edge soundblaster
332 8: 1 0 IO-APIC-edge rtc
333 9: 27422 27407 IO-APIC-edge 3c503
334 12: 113645 113873 IO-APIC-edge PS/2 Mouse
336 14: 22491 24012 IO-APIC-edge ide0
337 15: 2183 2415 IO-APIC-edge ide1
338 17: 30564 30414 IO-APIC-level eth0
339 18: 177 164 IO-APIC-level bttv
344 NMI is incremented in this case because every timer interrupt generates a NMI
345 (Non Maskable Interrupt) which is used by the NMI Watchdog to detect lockups.
347 LOC is the local interrupt counter of the internal APIC of every CPU.
349 ERR is incremented in the case of errors in the IO-APIC bus (the bus that
350 connects the CPUs in a SMP system. This means that an error has been detected,
351 the IO-APIC automatically retry the transmission, so it should not be a big
352 problem, but you should read the SMP-FAQ.
354 In 2.6.2* /proc/interrupts was expanded again. This time the goal was for
355 /proc/interrupts to display every IRQ vector in use by the system, not
356 just those considered 'most important'. The new vectors are:
358 THR -- interrupt raised when a machine check threshold counter
359 (typically counting ECC corrected errors of memory or cache) exceeds
360 a configurable threshold. Only available on some systems.
362 TRM -- a thermal event interrupt occurs when a temperature threshold
363 has been exceeded for the CPU. This interrupt may also be generated
364 when the temperature drops back to normal.
366 SPU -- a spurious interrupt is some interrupt that was raised then lowered
367 by some IO device before it could be fully processed by the APIC. Hence
368 the APIC sees the interrupt but does not know what device it came from.
369 For this case the APIC will generate the interrupt with a IRQ vector
370 of 0xff. This might also be generated by chipset bugs.
372 RES, CAL, TLB -- rescheduling, call and TLB flush interrupts are
373 sent from one CPU to another per the needs of the OS. Typically,
374 their statistics are used by kernel developers and interested users to
375 determine the occurance of interrupt of the given type.
377 The above IRQ vectors are displayed only when relevent. For example,
378 the threshold vector does not exist on x86_64 platforms. Others are
379 suppressed when the system is a uniprocessor. As of this writing, only
380 i386 and x86_64 platforms support the new IRQ vector displays.
382 Of some interest is the introduction of the /proc/irq directory to 2.4.
383 It could be used to set IRQ to CPU affinity, this means that you can "hook" an
384 IRQ to only one CPU, or to exclude a CPU of handling IRQs. The contents of the
385 irq subdir is one subdir for each IRQ, and two files; default_smp_affinity and
390 0 10 12 14 16 18 2 4 6 8 prof_cpu_mask
391 1 11 13 15 17 19 3 5 7 9 default_smp_affinity
395 smp_affinity is a bitmask, in which you can specify which CPUs can handle the
396 IRQ, you can set it by doing:
398 > echo 1 > /proc/irq/10/smp_affinity
400 This means that only the first CPU will handle the IRQ, but you can also echo
401 5 which means that only the first and fourth CPU can handle the IRQ.
403 The contents of each smp_affinity file is the same by default:
405 > cat /proc/irq/0/smp_affinity
408 The default_smp_affinity mask applies to all non-active IRQs, which are the
409 IRQs which have not yet been allocated/activated, and hence which lack a
410 /proc/irq/[0-9]* directory.
412 prof_cpu_mask specifies which CPUs are to be profiled by the system wide
413 profiler. Default value is ffffffff (all cpus).
415 The way IRQs are routed is handled by the IO-APIC, and it's Round Robin
416 between all the CPUs which are allowed to handle it. As usual the kernel has
417 more info than you and does a better job than you, so the defaults are the
418 best choice for almost everyone.
420 There are three more important subdirectories in /proc: net, scsi, and sys.
421 The general rule is that the contents, or even the existence of these
422 directories, depend on your kernel configuration. If SCSI is not enabled, the
423 directory scsi may not exist. The same is true with the net, which is there
424 only when networking support is present in the running kernel.
426 The slabinfo file gives information about memory usage at the slab level.
427 Linux uses slab pools for memory management above page level in version 2.2.
428 Commonly used objects have their own slab pool (such as network buffers,
429 directory cache, and so on).
431 ..............................................................................
433 > cat /proc/buddyinfo
435 Node 0, zone DMA 0 4 5 4 4 3 ...
436 Node 0, zone Normal 1 0 0 1 101 8 ...
437 Node 0, zone HighMem 2 0 0 1 1 0 ...
439 Memory fragmentation is a problem under some workloads, and buddyinfo is a
440 useful tool for helping diagnose these problems. Buddyinfo will give you a
441 clue as to how big an area you can safely allocate, or why a previous
444 Each column represents the number of pages of a certain order which are
445 available. In this case, there are 0 chunks of 2^0*PAGE_SIZE available in
446 ZONE_DMA, 4 chunks of 2^1*PAGE_SIZE in ZONE_DMA, 101 chunks of 2^4*PAGE_SIZE
447 available in ZONE_NORMAL, etc...
449 ..............................................................................
453 Provides information about distribution and utilization of memory. This
454 varies by architecture and compile options. The following is from a
455 16GB PIII, which has highmem enabled. You may not have all of these fields.
460 MemTotal: 16344972 kB
467 HighTotal: 15597528 kB
468 HighFree: 13629632 kB
478 SReclaimable: 159856 kB
479 SUnreclaim: 124508 kB
484 CommitLimit: 7669796 kB
485 Committed_AS: 100056 kB
486 VmallocTotal: 112216 kB
488 VmallocChunk: 111088 kB
490 MemTotal: Total usable ram (i.e. physical ram minus a few reserved
491 bits and the kernel binary code)
492 MemFree: The sum of LowFree+HighFree
493 Buffers: Relatively temporary storage for raw disk blocks
494 shouldn't get tremendously large (20MB or so)
495 Cached: in-memory cache for files read from the disk (the
496 pagecache). Doesn't include SwapCached
497 SwapCached: Memory that once was swapped out, is swapped back in but
498 still also is in the swapfile (if memory is needed it
499 doesn't need to be swapped out AGAIN because it is already
500 in the swapfile. This saves I/O)
501 Active: Memory that has been used more recently and usually not
502 reclaimed unless absolutely necessary.
503 Inactive: Memory which has been less recently used. It is more
504 eligible to be reclaimed for other purposes
506 HighFree: Highmem is all memory above ~860MB of physical memory
507 Highmem areas are for use by userspace programs, or
508 for the pagecache. The kernel must use tricks to access
509 this memory, making it slower to access than lowmem.
511 LowFree: Lowmem is memory which can be used for everything that
512 highmem can be used for, but it is also available for the
513 kernel's use for its own data structures. Among many
514 other things, it is where everything from the Slab is
515 allocated. Bad things happen when you're out of lowmem.
516 SwapTotal: total amount of swap space available
517 SwapFree: Memory which has been evicted from RAM, and is temporarily
519 Dirty: Memory which is waiting to get written back to the disk
520 Writeback: Memory which is actively being written back to the disk
521 AnonPages: Non-file backed pages mapped into userspace page tables
522 Mapped: files which have been mmaped, such as libraries
523 Slab: in-kernel data structures cache
524 SReclaimable: Part of Slab, that might be reclaimed, such as caches
525 SUnreclaim: Part of Slab, that cannot be reclaimed on memory pressure
526 PageTables: amount of memory dedicated to the lowest level of page
528 NFS_Unstable: NFS pages sent to the server, but not yet committed to stable
530 Bounce: Memory used for block device "bounce buffers"
531 WritebackTmp: Memory used by FUSE for temporary writeback buffers
532 CommitLimit: Based on the overcommit ratio ('vm.overcommit_ratio'),
533 this is the total amount of memory currently available to
534 be allocated on the system. This limit is only adhered to
535 if strict overcommit accounting is enabled (mode 2 in
536 'vm.overcommit_memory').
537 The CommitLimit is calculated with the following formula:
538 CommitLimit = ('vm.overcommit_ratio' * Physical RAM) + Swap
539 For example, on a system with 1G of physical RAM and 7G
540 of swap with a `vm.overcommit_ratio` of 30 it would
541 yield a CommitLimit of 7.3G.
542 For more details, see the memory overcommit documentation
543 in vm/overcommit-accounting.
544 Committed_AS: The amount of memory presently allocated on the system.
545 The committed memory is a sum of all of the memory which
546 has been allocated by processes, even if it has not been
547 "used" by them as of yet. A process which malloc()'s 1G
548 of memory, but only touches 300M of it will only show up
549 as using 300M of memory even if it has the address space
550 allocated for the entire 1G. This 1G is memory which has
551 been "committed" to by the VM and can be used at any time
552 by the allocating application. With strict overcommit
553 enabled on the system (mode 2 in 'vm.overcommit_memory'),
554 allocations which would exceed the CommitLimit (detailed
555 above) will not be permitted. This is useful if one needs
556 to guarantee that processes will not fail due to lack of
557 memory once that memory has been successfully allocated.
558 VmallocTotal: total size of vmalloc memory area
559 VmallocUsed: amount of vmalloc area which is used
560 VmallocChunk: largest contigious block of vmalloc area which is free
562 ..............................................................................
566 Provides information about vmalloced/vmaped areas. One line per area,
567 containing the virtual address range of the area, size in bytes,
568 caller information of the creator, and optional information depending
569 on the kind of area :
571 pages=nr number of pages
572 phys=addr if a physical address was specified
573 ioremap I/O mapping (ioremap() and friends)
574 vmalloc vmalloc() area
577 vpages buffer for pages pointers was vmalloced (huge area)
578 N<node>=nr (Only on NUMA kernels)
579 Number of pages allocated on memory node <node>
581 > cat /proc/vmallocinfo
582 0xffffc20000000000-0xffffc20000201000 2101248 alloc_large_system_hash+0x204 ...
583 /0x2c0 pages=512 vmalloc N0=128 N1=128 N2=128 N3=128
584 0xffffc20000201000-0xffffc20000302000 1052672 alloc_large_system_hash+0x204 ...
585 /0x2c0 pages=256 vmalloc N0=64 N1=64 N2=64 N3=64
586 0xffffc20000302000-0xffffc20000304000 8192 acpi_tb_verify_table+0x21/0x4f...
587 phys=7fee8000 ioremap
588 0xffffc20000304000-0xffffc20000307000 12288 acpi_tb_verify_table+0x21/0x4f...
589 phys=7fee7000 ioremap
590 0xffffc2000031d000-0xffffc2000031f000 8192 init_vdso_vars+0x112/0x210
591 0xffffc2000031f000-0xffffc2000032b000 49152 cramfs_uncompress_init+0x2e ...
592 /0x80 pages=11 vmalloc N0=3 N1=3 N2=2 N3=3
593 0xffffc2000033a000-0xffffc2000033d000 12288 sys_swapon+0x640/0xac0 ...
595 0xffffc20000347000-0xffffc2000034c000 20480 xt_alloc_table_info+0xfe ...
596 /0x130 [x_tables] pages=4 vmalloc N0=4
597 0xffffffffa0000000-0xffffffffa000f000 61440 sys_init_module+0xc27/0x1d00 ...
598 pages=14 vmalloc N2=14
599 0xffffffffa000f000-0xffffffffa0014000 20480 sys_init_module+0xc27/0x1d00 ...
601 0xffffffffa0014000-0xffffffffa0017000 12288 sys_init_module+0xc27/0x1d00 ...
603 0xffffffffa0017000-0xffffffffa0022000 45056 sys_init_module+0xc27/0x1d00 ...
604 pages=10 vmalloc N0=10
606 1.3 IDE devices in /proc/ide
607 ----------------------------
609 The subdirectory /proc/ide contains information about all IDE devices of which
610 the kernel is aware. There is one subdirectory for each IDE controller, the
611 file drivers and a link for each IDE device, pointing to the device directory
612 in the controller specific subtree.
614 The file drivers contains general information about the drivers used for the
617 > cat /proc/ide/drivers
618 ide-cdrom version 4.53
619 ide-disk version 1.08
621 More detailed information can be found in the controller specific
622 subdirectories. These are named ide0, ide1 and so on. Each of these
623 directories contains the files shown in table 1-5.
626 Table 1-5: IDE controller info in /proc/ide/ide?
627 ..............................................................................
629 channel IDE channel (0 or 1)
630 config Configuration (only for PCI/IDE bridge)
632 model Type/Chipset of IDE controller
633 ..............................................................................
635 Each device connected to a controller has a separate subdirectory in the
636 controllers directory. The files listed in table 1-6 are contained in these
640 Table 1-6: IDE device information
641 ..............................................................................
644 capacity Capacity of the medium (in 512Byte blocks)
645 driver driver and version
646 geometry physical and logical geometry
647 identify device identify block
649 model device identifier
650 settings device setup
651 smart_thresholds IDE disk management thresholds
652 smart_values IDE disk management values
653 ..............................................................................
655 The most interesting file is settings. This file contains a nice overview of
656 the drive parameters:
658 # cat /proc/ide/ide0/hda/settings
659 name value min max mode
660 ---- ----- --- --- ----
661 bios_cyl 526 0 65535 rw
662 bios_head 255 0 255 rw
664 breada_readahead 4 0 127 rw
666 file_readahead 72 0 2097151 rw
668 keepsettings 0 0 1 rw
669 max_kb_per_request 122 1 127 rw
673 pio_mode write-only 0 255 w
679 1.4 Networking info in /proc/net
680 --------------------------------
682 The subdirectory /proc/net follows the usual pattern. Table 1-6 shows the
683 additional values you get for IP version 6 if you configure the kernel to
684 support this. Table 1-7 lists the files and their meaning.
687 Table 1-6: IPv6 info in /proc/net
688 ..............................................................................
690 udp6 UDP sockets (IPv6)
691 tcp6 TCP sockets (IPv6)
692 raw6 Raw device statistics (IPv6)
693 igmp6 IP multicast addresses, which this host joined (IPv6)
694 if_inet6 List of IPv6 interface addresses
695 ipv6_route Kernel routing table for IPv6
696 rt6_stats Global IPv6 routing tables statistics
697 sockstat6 Socket statistics (IPv6)
698 snmp6 Snmp data (IPv6)
699 ..............................................................................
702 Table 1-7: Network info in /proc/net
703 ..............................................................................
706 dev network devices with statistics
707 dev_mcast the Layer2 multicast groups a device is listening too
708 (interface index, label, number of references, number of bound
710 dev_stat network device status
711 ip_fwchains Firewall chain linkage
712 ip_fwnames Firewall chain names
713 ip_masq Directory containing the masquerading tables
714 ip_masquerade Major masquerading table
715 netstat Network statistics
716 raw raw device statistics
717 route Kernel routing table
718 rpc Directory containing rpc info
719 rt_cache Routing cache
721 sockstat Socket statistics
723 tr_rif Token ring RIF routing table
725 unix UNIX domain sockets
726 wireless Wireless interface data (Wavelan etc)
727 igmp IP multicast addresses, which this host joined
728 psched Global packet scheduler parameters.
729 netlink List of PF_NETLINK sockets
730 ip_mr_vifs List of multicast virtual interfaces
731 ip_mr_cache List of multicast routing cache
732 ..............................................................................
734 You can use this information to see which network devices are available in
735 your system and how much traffic was routed over those devices:
739 face |bytes packets errs drop fifo frame compressed multicast|[...
740 lo: 908188 5596 0 0 0 0 0 0 [...
741 ppp0:15475140 20721 410 0 0 410 0 0 [...
742 eth0: 614530 7085 0 0 0 0 0 1 [...
745 ...] bytes packets errs drop fifo colls carrier compressed
746 ...] 908188 5596 0 0 0 0 0 0
747 ...] 1375103 17405 0 0 0 0 0 0
748 ...] 1703981 5535 0 0 0 3 0 0
750 In addition, each Channel Bond interface has it's own directory. For
751 example, the bond0 device will have a directory called /proc/net/bond0/.
752 It will contain information that is specific to that bond, such as the
753 current slaves of the bond, the link status of the slaves, and how
754 many times the slaves link has failed.
759 If you have a SCSI host adapter in your system, you'll find a subdirectory
760 named after the driver for this adapter in /proc/scsi. You'll also see a list
761 of all recognized SCSI devices in /proc/scsi:
765 Host: scsi0 Channel: 00 Id: 00 Lun: 00
766 Vendor: IBM Model: DGHS09U Rev: 03E0
767 Type: Direct-Access ANSI SCSI revision: 03
768 Host: scsi0 Channel: 00 Id: 06 Lun: 00
769 Vendor: PIONEER Model: CD-ROM DR-U06S Rev: 1.04
770 Type: CD-ROM ANSI SCSI revision: 02
773 The directory named after the driver has one file for each adapter found in
774 the system. These files contain information about the controller, including
775 the used IRQ and the IO address range. The amount of information shown is
776 dependent on the adapter you use. The example shows the output for an Adaptec
777 AHA-2940 SCSI adapter:
779 > cat /proc/scsi/aic7xxx/0
781 Adaptec AIC7xxx driver version: 5.1.19/3.2.4
783 TCQ Enabled By Default : Disabled
784 AIC7XXX_PROC_STATS : Disabled
785 AIC7XXX_RESET_DELAY : 5
786 Adapter Configuration:
787 SCSI Adapter: Adaptec AHA-294X Ultra SCSI host adapter
788 Ultra Wide Controller
789 PCI MMAPed I/O Base: 0xeb001000
790 Adapter SEEPROM Config: SEEPROM found and used.
791 Adaptec SCSI BIOS: Enabled
793 SCBs: Active 0, Max Active 2,
794 Allocated 15, HW 16, Page 255
796 BIOS Control Word: 0x18b6
797 Adapter Control Word: 0x005b
798 Extended Translation: Enabled
799 Disconnect Enable Flags: 0xffff
800 Ultra Enable Flags: 0x0001
801 Tag Queue Enable Flags: 0x0000
802 Ordered Queue Tag Flags: 0x0000
803 Default Tag Queue Depth: 8
804 Tagged Queue By Device array for aic7xxx host instance 0:
805 {255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255}
806 Actual queue depth per device for aic7xxx host instance 0:
807 {1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1}
810 Device using Wide/Sync transfers at 40.0 MByte/sec, offset 8
811 Transinfo settings: current(12/8/1/0), goal(12/8/1/0), user(12/15/1/0)
812 Total transfers 160151 (74577 reads and 85574 writes)
814 Device using Narrow/Sync transfers at 5.0 MByte/sec, offset 15
815 Transinfo settings: current(50/15/0/0), goal(50/15/0/0), user(50/15/0/0)
816 Total transfers 0 (0 reads and 0 writes)
819 1.6 Parallel port info in /proc/parport
820 ---------------------------------------
822 The directory /proc/parport contains information about the parallel ports of
823 your system. It has one subdirectory for each port, named after the port
826 These directories contain the four files shown in Table 1-8.
829 Table 1-8: Files in /proc/parport
830 ..............................................................................
832 autoprobe Any IEEE-1284 device ID information that has been acquired.
833 devices list of the device drivers using that port. A + will appear by the
834 name of the device currently using the port (it might not appear
836 hardware Parallel port's base address, IRQ line and DMA channel.
837 irq IRQ that parport is using for that port. This is in a separate
838 file to allow you to alter it by writing a new value in (IRQ
840 ..............................................................................
842 1.7 TTY info in /proc/tty
843 -------------------------
845 Information about the available and actually used tty's can be found in the
846 directory /proc/tty.You'll find entries for drivers and line disciplines in
847 this directory, as shown in Table 1-9.
850 Table 1-9: Files in /proc/tty
851 ..............................................................................
853 drivers list of drivers and their usage
854 ldiscs registered line disciplines
855 driver/serial usage statistic and status of single tty lines
856 ..............................................................................
858 To see which tty's are currently in use, you can simply look into the file
861 > cat /proc/tty/drivers
862 pty_slave /dev/pts 136 0-255 pty:slave
863 pty_master /dev/ptm 128 0-255 pty:master
864 pty_slave /dev/ttyp 3 0-255 pty:slave
865 pty_master /dev/pty 2 0-255 pty:master
866 serial /dev/cua 5 64-67 serial:callout
867 serial /dev/ttyS 4 64-67 serial
868 /dev/tty0 /dev/tty0 4 0 system:vtmaster
869 /dev/ptmx /dev/ptmx 5 2 system
870 /dev/console /dev/console 5 1 system:console
871 /dev/tty /dev/tty 5 0 system:/dev/tty
872 unknown /dev/tty 4 1-63 console
875 1.8 Miscellaneous kernel statistics in /proc/stat
876 -------------------------------------------------
878 Various pieces of information about kernel activity are available in the
879 /proc/stat file. All of the numbers reported in this file are aggregates
880 since the system first booted. For a quick look, simply cat the file:
883 cpu 2255 34 2290 22625563 6290 127 456 0
884 cpu0 1132 34 1441 11311718 3675 127 438 0
885 cpu1 1123 0 849 11313845 2614 0 18 0
886 intr 114930548 113199788 3 0 5 263 0 4 [... lots more numbers ...]
893 The very first "cpu" line aggregates the numbers in all of the other "cpuN"
894 lines. These numbers identify the amount of time the CPU has spent performing
895 different kinds of work. Time units are in USER_HZ (typically hundredths of a
896 second). The meanings of the columns are as follows, from left to right:
898 - user: normal processes executing in user mode
899 - nice: niced processes executing in user mode
900 - system: processes executing in kernel mode
901 - idle: twiddling thumbs
902 - iowait: waiting for I/O to complete
903 - irq: servicing interrupts
904 - softirq: servicing softirqs
905 - steal: involuntary wait
907 The "intr" line gives counts of interrupts serviced since boot time, for each
908 of the possible system interrupts. The first column is the total of all
909 interrupts serviced; each subsequent column is the total for that particular
912 The "ctxt" line gives the total number of context switches across all CPUs.
914 The "btime" line gives the time at which the system booted, in seconds since
917 The "processes" line gives the number of processes and threads created, which
918 includes (but is not limited to) those created by calls to the fork() and
919 clone() system calls.
921 The "procs_running" line gives the number of processes currently running on
924 The "procs_blocked" line gives the number of processes currently blocked,
925 waiting for I/O to complete.
928 1.9 Ext4 file system parameters
929 ------------------------------
931 Information about mounted ext4 file systems can be found in
932 /proc/fs/ext4. Each mounted filesystem will have a directory in
933 /proc/fs/ext4 based on its device name (i.e., /proc/fs/ext4/hdc or
934 /proc/fs/ext4/dm-0). The files in each per-device directory are shown
935 in Table 1-10, below.
937 Table 1-10: Files in /proc/fs/ext4/<devname>
938 ..............................................................................
940 mb_groups details of multiblock allocator buddy cache of free blocks
941 mb_history multiblock allocation history
942 stats controls whether the multiblock allocator should start
943 collecting statistics, which are shown during the unmount
944 group_prealloc the multiblock allocator will round up allocation
945 requests to a multiple of this tuning parameter if the
946 stripe size is not set in the ext4 superblock
947 max_to_scan The maximum number of extents the multiblock allocator
948 will search to find the best extent
949 min_to_scan The minimum number of extents the multiblock allocator
950 will search to find the best extent
951 order2_req Tuning parameter which controls the minimum size for
952 requests (as a power of 2) where the buddy cache is
954 stream_req Files which have fewer blocks than this tunable
955 parameter will have their blocks allocated out of a
956 block group specific preallocation pool, so that small
957 files are packed closely together. Each large file
958 will have its blocks allocated out of its own unique
960 inode_readahead Tuning parameter which controls the maximum number of
961 inode table blocks that ext4's inode table readahead
962 algorithm will pre-read into the buffer cache
963 ..............................................................................
966 ------------------------------------------------------------------------------
968 ------------------------------------------------------------------------------
969 The /proc file system serves information about the running system. It not only
970 allows access to process data but also allows you to request the kernel status
971 by reading files in the hierarchy.
973 The directory structure of /proc reflects the types of information and makes
974 it easy, if not obvious, where to look for specific data.
975 ------------------------------------------------------------------------------
977 ------------------------------------------------------------------------------
978 CHAPTER 2: MODIFYING SYSTEM PARAMETERS
979 ------------------------------------------------------------------------------
981 ------------------------------------------------------------------------------
983 ------------------------------------------------------------------------------
984 * Modifying kernel parameters by writing into files found in /proc/sys
985 * Exploring the files which modify certain parameters
986 * Review of the /proc/sys file tree
987 ------------------------------------------------------------------------------
990 A very interesting part of /proc is the directory /proc/sys. This is not only
991 a source of information, it also allows you to change parameters within the
992 kernel. Be very careful when attempting this. You can optimize your system,
993 but you can also cause it to crash. Never alter kernel parameters on a
994 production system. Set up a development machine and test to make sure that
995 everything works the way you want it to. You may have no alternative but to
996 reboot the machine once an error has been made.
998 To change a value, simply echo the new value into the file. An example is
999 given below in the section on the file system data. You need to be root to do
1000 this. You can create your own boot script to perform this every time your
1003 The files in /proc/sys can be used to fine tune and monitor miscellaneous and
1004 general things in the operation of the Linux kernel. Since some of the files
1005 can inadvertently disrupt your system, it is advisable to read both
1006 documentation and source before actually making adjustments. In any case, be
1007 very careful when writing to any of these files. The entries in /proc may
1008 change slightly between the 2.1.* and the 2.2 kernel, so if there is any doubt
1009 review the kernel documentation in the directory /usr/src/linux/Documentation.
1010 This chapter is heavily based on the documentation included in the pre 2.2
1011 kernels, and became part of it in version 2.2.1 of the Linux kernel.
1013 2.1 /proc/sys/fs - File system data
1014 -----------------------------------
1016 This subdirectory contains specific file system, file handle, inode, dentry
1017 and quota information.
1019 Currently, these files are in /proc/sys/fs:
1024 Status of the directory cache. Since directory entries are dynamically
1025 allocated and deallocated, this file indicates the current status. It holds
1026 six values, in which the last two are not used and are always zero. The others
1027 are listed in table 2-1.
1030 Table 2-1: Status files of the directory cache
1031 ..............................................................................
1033 nr_dentry Almost always zero
1034 nr_unused Number of unused cache entries
1036 in seconds after the entry may be reclaimed, when memory is short
1037 want_pages internally
1038 ..............................................................................
1040 dquot-nr and dquot-max
1041 ----------------------
1043 The file dquot-max shows the maximum number of cached disk quota entries.
1045 The file dquot-nr shows the number of allocated disk quota entries and the
1046 number of free disk quota entries.
1048 If the number of available cached disk quotas is very low and you have a large
1049 number of simultaneous system users, you might want to raise the limit.
1051 file-nr and file-max
1052 --------------------
1054 The kernel allocates file handles dynamically, but doesn't free them again at
1057 The value in file-max denotes the maximum number of file handles that the
1058 Linux kernel will allocate. When you get a lot of error messages about running
1059 out of file handles, you might want to raise this limit. The default value is
1060 10% of RAM in kilobytes. To change it, just write the new number into the
1063 # cat /proc/sys/fs/file-max
1065 # echo 8192 > /proc/sys/fs/file-max
1066 # cat /proc/sys/fs/file-max
1070 This method of revision is useful for all customizable parameters of the
1071 kernel - simply echo the new value to the corresponding file.
1073 Historically, the three values in file-nr denoted the number of allocated file
1074 handles, the number of allocated but unused file handles, and the maximum
1075 number of file handles. Linux 2.6 always reports 0 as the number of free file
1076 handles -- this is not an error, it just means that the number of allocated
1077 file handles exactly matches the number of used file handles.
1079 Attempts to allocate more file descriptors than file-max are reported with
1080 printk, look for "VFS: file-max limit <number> reached".
1082 inode-state and inode-nr
1083 ------------------------
1085 The file inode-nr contains the first two items from inode-state, so we'll skip
1088 inode-state contains two actual numbers and five dummy values. The numbers
1089 are nr_inodes and nr_free_inodes (in order of appearance).
1094 Denotes the number of inodes the system has allocated. This number will
1095 grow and shrink dynamically.
1100 Denotes the maximum number of file-handles a process can
1101 allocate. Default value is 1024*1024 (1048576) which should be
1102 enough for most machines. Actual limit depends on RLIMIT_NOFILE
1108 Represents the number of free inodes. Ie. The number of inuse inodes is
1109 (nr_inodes - nr_free_inodes).
1111 aio-nr and aio-max-nr
1112 ---------------------
1114 aio-nr is the running total of the number of events specified on the
1115 io_setup system call for all currently active aio contexts. If aio-nr
1116 reaches aio-max-nr then io_setup will fail with EAGAIN. Note that
1117 raising aio-max-nr does not result in the pre-allocation or re-sizing
1118 of any kernel data structures.
1120 2.2 /proc/sys/fs/binfmt_misc - Miscellaneous binary formats
1121 -----------------------------------------------------------
1123 Besides these files, there is the subdirectory /proc/sys/fs/binfmt_misc. This
1124 handles the kernel support for miscellaneous binary formats.
1126 Binfmt_misc provides the ability to register additional binary formats to the
1127 Kernel without compiling an additional module/kernel. Therefore, binfmt_misc
1128 needs to know magic numbers at the beginning or the filename extension of the
1131 It works by maintaining a linked list of structs that contain a description of
1132 a binary format, including a magic with size (or the filename extension),
1133 offset and mask, and the interpreter name. On request it invokes the given
1134 interpreter with the original program as argument, as binfmt_java and
1135 binfmt_em86 and binfmt_mz do. Since binfmt_misc does not define any default
1136 binary-formats, you have to register an additional binary-format.
1138 There are two general files in binfmt_misc and one file per registered format.
1139 The two general files are register and status.
1141 Registering a new binary format
1142 -------------------------------
1144 To register a new binary format you have to issue the command
1146 echo :name:type:offset:magic:mask:interpreter: > /proc/sys/fs/binfmt_misc/register
1150 with appropriate name (the name for the /proc-dir entry), offset (defaults to
1151 0, if omitted), magic, mask (which can be omitted, defaults to all 0xff) and
1152 last but not least, the interpreter that is to be invoked (for example and
1153 testing /bin/echo). Type can be M for usual magic matching or E for filename
1154 extension matching (give extension in place of magic).
1156 Check or reset the status of the binary format handler
1157 ------------------------------------------------------
1159 If you do a cat on the file /proc/sys/fs/binfmt_misc/status, you will get the
1160 current status (enabled/disabled) of binfmt_misc. Change the status by echoing
1161 0 (disables) or 1 (enables) or -1 (caution: this clears all previously
1162 registered binary formats) to status. For example echo 0 > status to disable
1163 binfmt_misc (temporarily).
1165 Status of a single handler
1166 --------------------------
1168 Each registered handler has an entry in /proc/sys/fs/binfmt_misc. These files
1169 perform the same function as status, but their scope is limited to the actual
1170 binary format. By cating this file, you also receive all related information
1171 about the interpreter/magic of the binfmt.
1173 Example usage of binfmt_misc (emulate binfmt_java)
1174 --------------------------------------------------
1176 cd /proc/sys/fs/binfmt_misc
1177 echo ':Java:M::\xca\xfe\xba\xbe::/usr/local/java/bin/javawrapper:' > register
1178 echo ':HTML:E::html::/usr/local/java/bin/appletviewer:' > register
1179 echo ':Applet:M::<!--applet::/usr/local/java/bin/appletviewer:' > register
1180 echo ':DEXE:M::\x0eDEX::/usr/bin/dosexec:' > register
1183 These four lines add support for Java executables and Java applets (like
1184 binfmt_java, additionally recognizing the .html extension with no need to put
1185 <!--applet> to every applet file). You have to install the JDK and the
1186 shell-script /usr/local/java/bin/javawrapper too. It works around the
1187 brokenness of the Java filename handling. To add a Java binary, just create a
1188 link to the class-file somewhere in the path.
1190 2.3 /proc/sys/kernel - general kernel parameters
1191 ------------------------------------------------
1193 This directory reflects general kernel behaviors. As I've said before, the
1194 contents depend on your configuration. Here you'll find the most important
1195 files, along with descriptions of what they mean and how to use them.
1200 The file contains three values; highwater, lowwater, and frequency.
1202 It exists only when BSD-style process accounting is enabled. These values
1203 control its behavior. If the free space on the file system where the log lives
1204 goes below lowwater percentage, accounting suspends. If it goes above
1205 highwater percentage, accounting resumes. Frequency determines how often you
1206 check the amount of free space (value is in seconds). Default settings are: 4,
1207 2, and 30. That is, suspend accounting if there is less than 2 percent free;
1208 resume it if we have a value of 3 or more percent; consider information about
1209 the amount of free space valid for 30 seconds
1214 When the value in this file is 0, ctrl-alt-del is trapped and sent to the init
1215 program to handle a graceful restart. However, when the value is greater that
1216 zero, Linux's reaction to this key combination will be an immediate reboot,
1217 without syncing its dirty buffers.
1220 When a program (like dosemu) has the keyboard in raw mode, the
1221 ctrl-alt-del is intercepted by the program before it ever reaches the
1222 kernel tty layer, and it is up to the program to decide what to do with
1225 domainname and hostname
1226 -----------------------
1228 These files can be controlled to set the NIS domainname and hostname of your
1229 box. For the classic darkstar.frop.org a simple:
1231 # echo "darkstar" > /proc/sys/kernel/hostname
1232 # echo "frop.org" > /proc/sys/kernel/domainname
1235 would suffice to set your hostname and NIS domainname.
1237 osrelease, ostype and version
1238 -----------------------------
1240 The names make it pretty obvious what these fields contain:
1242 > cat /proc/sys/kernel/osrelease
1245 > cat /proc/sys/kernel/ostype
1248 > cat /proc/sys/kernel/version
1249 #4 Fri Oct 1 12:41:14 PDT 1999
1252 The files osrelease and ostype should be clear enough. Version needs a little
1253 more clarification. The #4 means that this is the 4th kernel built from this
1254 source base and the date after it indicates the time the kernel was built. The
1255 only way to tune these values is to rebuild the kernel.
1260 The value in this file represents the number of seconds the kernel waits
1261 before rebooting on a panic. When you use the software watchdog, the
1262 recommended setting is 60. If set to 0, the auto reboot after a kernel panic
1263 is disabled, which is the default setting.
1268 The four values in printk denote
1270 * default_message_loglevel,
1271 * minimum_console_loglevel and
1272 * default_console_loglevel
1275 These values influence printk() behavior when printing or logging error
1276 messages, which come from inside the kernel. See syslog(2) for more
1277 information on the different log levels.
1282 Messages with a higher priority than this will be printed to the console.
1284 default_message_level
1285 ---------------------
1287 Messages without an explicit priority will be printed with this priority.
1289 minimum_console_loglevel
1290 ------------------------
1292 Minimum (highest) value to which the console_loglevel can be set.
1294 default_console_loglevel
1295 ------------------------
1297 Default value for console_loglevel.
1302 This file shows the size of the generic SCSI (sg) buffer. At this point, you
1303 can't tune it yet, but you can change it at compile time by editing
1304 include/scsi/sg.h and changing the value of SG_BIG_BUFF.
1306 If you use a scanner with SANE (Scanner Access Now Easy) you might want to set
1307 this to a higher value. Refer to the SANE documentation on this issue.
1312 The location where the modprobe binary is located. The kernel uses this
1313 program to load modules on demand.
1318 The value in this file affects behavior of handling NMI. When the value is
1319 non-zero, unknown NMI is trapped and then panic occurs. At that time, kernel
1320 debugging information is displayed on console.
1322 NMI switch that most IA32 servers have fires unknown NMI up, for example.
1323 If a system hangs up, try pressing the NMI switch.
1325 panic_on_unrecovered_nmi
1326 ------------------------
1328 The default Linux behaviour on an NMI of either memory or unknown is to continue
1329 operation. For many environments such as scientific computing it is preferable
1330 that the box is taken out and the error dealt with than an uncorrected
1331 parity/ECC error get propogated.
1333 A small number of systems do generate NMI's for bizarre random reasons such as
1334 power management so the default is off. That sysctl works like the existing
1335 panic controls already in that directory.
1340 Enables/Disables the NMI watchdog on x86 systems. When the value is non-zero
1341 the NMI watchdog is enabled and will continuously test all online cpus to
1342 determine whether or not they are still functioning properly. Currently,
1343 passing "nmi_watchdog=" parameter at boot time is required for this function
1346 If LAPIC NMI watchdog method is in use (nmi_watchdog=2 kernel parameter), the
1347 NMI watchdog shares registers with oprofile. By disabling the NMI watchdog,
1348 oprofile may have more registers to utilize.
1353 Maximum number of message queue ids on the system.
1354 This value scales to the amount of lowmem. It is automatically recomputed
1355 upon memory add/remove or ipc namespace creation/removal.
1356 When a value is written into this file, msgmni's value becomes fixed, i.e. it
1357 is not recomputed anymore when one of the above events occurs.
1358 Use auto_msgmni to change this behavior.
1363 Enables/Disables automatic recomputing of msgmni upon memory add/remove or
1364 upon ipc namespace creation/removal (see the msgmni description above).
1365 Echoing "1" into this file enables msgmni automatic recomputing.
1366 Echoing "0" turns it off.
1367 auto_msgmni default value is 1.
1370 2.4 /proc/sys/vm - The virtual memory subsystem
1371 -----------------------------------------------
1373 The files in this directory can be used to tune the operation of the virtual
1374 memory (VM) subsystem of the Linux kernel.
1379 Controls the tendency of the kernel to reclaim the memory which is used for
1380 caching of directory and inode objects.
1382 At the default value of vfs_cache_pressure=100 the kernel will attempt to
1383 reclaim dentries and inodes at a "fair" rate with respect to pagecache and
1384 swapcache reclaim. Decreasing vfs_cache_pressure causes the kernel to prefer
1385 to retain dentry and inode caches. Increasing vfs_cache_pressure beyond 100
1386 causes the kernel to prefer to reclaim dentries and inodes.
1388 dirty_background_ratio
1389 ----------------------
1391 Contains, as a percentage of the dirtyable system memory (free pages + mapped
1392 pages + file cache, not including locked pages and HugePages), the number of
1393 pages at which the pdflush background writeback daemon will start writing out
1399 Contains, as a percentage of the dirtyable system memory (free pages + mapped
1400 pages + file cache, not including locked pages and HugePages), the number of
1401 pages at which a process which is generating disk writes will itself start
1402 writing out dirty data.
1404 dirty_writeback_centisecs
1405 -------------------------
1407 The pdflush writeback daemons will periodically wake up and write `old' data
1408 out to disk. This tunable expresses the interval between those wakeups, in
1409 100'ths of a second.
1411 Setting this to zero disables periodic writeback altogether.
1413 dirty_expire_centisecs
1414 ----------------------
1416 This tunable is used to define when dirty data is old enough to be eligible
1417 for writeout by the pdflush daemons. It is expressed in 100'ths of a second.
1418 Data which has been dirty in-memory for longer than this interval will be
1419 written out next time a pdflush daemon wakes up.
1421 highmem_is_dirtyable
1422 --------------------
1424 Only present if CONFIG_HIGHMEM is set.
1426 This defaults to 0 (false), meaning that the ratios set above are calculated
1427 as a percentage of lowmem only. This protects against excessive scanning
1428 in page reclaim, swapping and general VM distress.
1430 Setting this to 1 can be useful on 32 bit machines where you want to make
1431 random changes within an MMAPed file that is larger than your available
1432 lowmem without causing large quantities of random IO. Is is safe if the
1433 behavior of all programs running on the machine is known and memory will
1434 not be otherwise stressed.
1439 If non-zero, this sysctl disables the new 32-bit mmap mmap layout - the kernel
1440 will use the legacy (2.4) layout for all processes.
1442 lowmem_reserve_ratio
1443 ---------------------
1445 For some specialised workloads on highmem machines it is dangerous for
1446 the kernel to allow process memory to be allocated from the "lowmem"
1447 zone. This is because that memory could then be pinned via the mlock()
1448 system call, or by unavailability of swapspace.
1450 And on large highmem machines this lack of reclaimable lowmem memory
1453 So the Linux page allocator has a mechanism which prevents allocations
1454 which _could_ use highmem from using too much lowmem. This means that
1455 a certain amount of lowmem is defended from the possibility of being
1456 captured into pinned user memory.
1458 (The same argument applies to the old 16 megabyte ISA DMA region. This
1459 mechanism will also defend that region from allocations which could use
1462 The `lowmem_reserve_ratio' tunable determines how aggressive the kernel is
1463 in defending these lower zones.
1465 If you have a machine which uses highmem or ISA DMA and your
1466 applications are using mlock(), or if you are running with no swap then
1467 you probably should change the lowmem_reserve_ratio setting.
1469 The lowmem_reserve_ratio is an array. You can see them by reading this file.
1471 % cat /proc/sys/vm/lowmem_reserve_ratio
1474 Note: # of this elements is one fewer than number of zones. Because the highest
1475 zone's value is not necessary for following calculation.
1477 But, these values are not used directly. The kernel calculates # of protection
1478 pages for each zones from them. These are shown as array of protection pages
1479 in /proc/zoneinfo like followings. (This is an example of x86-64 box).
1480 Each zone has an array of protection pages like this.
1491 protection: (0, 2004, 2004, 2004)
1492 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
1497 These protections are added to score to judge whether this zone should be used
1498 for page allocation or should be reclaimed.
1500 In this example, if normal pages (index=2) are required to this DMA zone and
1501 pages_high is used for watermark, the kernel judges this zone should not be
1502 used because pages_free(1355) is smaller than watermark + protection[2]
1503 (4 + 2004 = 2008). If this protection value is 0, this zone would be used for
1504 normal page requirement. If requirement is DMA zone(index=0), protection[0]
1507 zone[i]'s protection[j] is calculated by following expression.
1510 zone[i]->protection[j]
1511 = (total sums of present_pages from zone[i+1] to zone[j] on the node)
1512 / lowmem_reserve_ratio[i];
1514 (should not be protected. = 0;
1516 (not necessary, but looks 0)
1518 The default values of lowmem_reserve_ratio[i] are
1519 256 (if zone[i] means DMA or DMA32 zone)
1521 As above expression, they are reciprocal number of ratio.
1522 256 means 1/256. # of protection pages becomes about "0.39%" of total present
1523 pages of higher zones on the node.
1525 If you would like to protect more pages, smaller values are effective.
1526 The minimum value is 1 (1/1 -> 100%).
1531 page-cluster controls the number of pages which are written to swap in
1532 a single attempt. The swap I/O size.
1534 It is a logarithmic value - setting it to zero means "1 page", setting
1535 it to 1 means "2 pages", setting it to 2 means "4 pages", etc.
1537 The default value is three (eight pages at a time). There may be some
1538 small benefits in tuning this to a different value if your workload is
1544 Controls overcommit of system memory, possibly allowing processes
1545 to allocate (but not use) more memory than is actually available.
1548 0 - Heuristic overcommit handling. Obvious overcommits of
1549 address space are refused. Used for a typical system. It
1550 ensures a seriously wild allocation fails while allowing
1551 overcommit to reduce swap usage. root is allowed to
1552 allocate slightly more memory in this mode. This is the
1555 1 - Always overcommit. Appropriate for some scientific
1558 2 - Don't overcommit. The total address space commit
1559 for the system is not permitted to exceed swap plus a
1560 configurable percentage (default is 50) of physical RAM.
1561 Depending on the percentage you use, in most situations
1562 this means a process will not be killed while attempting
1563 to use already-allocated memory but will receive errors
1564 on memory allocation as appropriate.
1569 Percentage of physical memory size to include in overcommit calculations
1572 Memory allocation limit = swapspace + physmem * (overcommit_ratio / 100)
1574 swapspace = total size of all swap areas
1575 physmem = size of physical memory in system
1577 nr_hugepages and hugetlb_shm_group
1578 ----------------------------------
1580 nr_hugepages configures number of hugetlb page reserved for the system.
1582 hugetlb_shm_group contains group id that is allowed to create SysV shared
1583 memory segment using hugetlb page.
1585 hugepages_treat_as_movable
1586 --------------------------
1588 This parameter is only useful when kernelcore= is specified at boot time to
1589 create ZONE_MOVABLE for pages that may be reclaimed or migrated. Huge pages
1590 are not movable so are not normally allocated from ZONE_MOVABLE. A non-zero
1591 value written to hugepages_treat_as_movable allows huge pages to be allocated
1594 Once enabled, the ZONE_MOVABLE is treated as an area of memory the huge
1595 pages pool can easily grow or shrink within. Assuming that applications are
1596 not running that mlock() a lot of memory, it is likely the huge pages pool
1597 can grow to the size of ZONE_MOVABLE by repeatedly entering the desired value
1598 into nr_hugepages and triggering page reclaim.
1603 laptop_mode is a knob that controls "laptop mode". All the things that are
1604 controlled by this knob are discussed in Documentation/laptops/laptop-mode.txt.
1609 block_dump enables block I/O debugging when set to a nonzero value. More
1610 information on block I/O debugging is in Documentation/laptops/laptop-mode.txt.
1615 This file contains valid hold time of swap out protection token. The Linux
1616 VM has token based thrashing control mechanism and uses the token to prevent
1617 unnecessary page faults in thrashing situation. The unit of the value is
1618 second. The value would be useful to tune thrashing behavior.
1623 Writing to this will cause the kernel to drop clean caches, dentries and
1624 inodes from memory, causing that memory to become free.
1627 echo 1 > /proc/sys/vm/drop_caches
1628 To free dentries and inodes:
1629 echo 2 > /proc/sys/vm/drop_caches
1630 To free pagecache, dentries and inodes:
1631 echo 3 > /proc/sys/vm/drop_caches
1633 As this is a non-destructive operation and dirty objects are not freeable, the
1634 user should run `sync' first.
1637 2.5 /proc/sys/dev - Device specific parameters
1638 ----------------------------------------------
1640 Currently there is only support for CDROM drives, and for those, there is only
1641 one read-only file containing information about the CD-ROM drives attached to
1644 >cat /proc/sys/dev/cdrom/info
1645 CD-ROM information, Id: cdrom.c 2.55 1999/04/25
1649 drive # of slots: 1 0
1653 Can change speed: 1 1
1654 Can select disk: 0 1
1655 Can read multisession: 1 1
1657 Reports media changed: 1 1
1661 You see two drives, sr0 and hdb, along with a list of their features.
1663 2.6 /proc/sys/sunrpc - Remote procedure calls
1664 ---------------------------------------------
1666 This directory contains four files, which enable or disable debugging for the
1667 RPC functions NFS, NFS-daemon, RPC and NLM. The default values are 0. They can
1668 be set to one to turn debugging on. (The default value is 0 for each)
1670 2.7 /proc/sys/net - Networking stuff
1671 ------------------------------------
1673 The interface to the networking parts of the kernel is located in
1674 /proc/sys/net. Table 2-3 shows all possible subdirectories. You may see only
1675 some of them, depending on your kernel's configuration.
1678 Table 2-3: Subdirectories in /proc/sys/net
1679 ..............................................................................
1680 Directory Content Directory Content
1681 core General parameter appletalk Appletalk protocol
1682 unix Unix domain sockets netrom NET/ROM
1683 802 E802 protocol ax25 AX25
1684 ethernet Ethernet protocol rose X.25 PLP layer
1685 ipv4 IP version 4 x25 X.25 protocol
1686 ipx IPX token-ring IBM token ring
1687 bridge Bridging decnet DEC net
1689 ..............................................................................
1691 We will concentrate on IP networking here. Since AX15, X.25, and DEC Net are
1692 only minor players in the Linux world, we'll skip them in this chapter. You'll
1693 find some short info on Appletalk and IPX further on in this chapter. Review
1694 the online documentation and the kernel source to get a detailed view of the
1695 parameters for those protocols. In this section we'll discuss the
1696 subdirectories printed in bold letters in the table above. As default values
1697 are suitable for most needs, there is no need to change these values.
1699 /proc/sys/net/core - Network core options
1700 -----------------------------------------
1705 The default setting of the socket receive buffer in bytes.
1710 The maximum receive socket buffer size in bytes.
1715 The default setting (in bytes) of the socket send buffer.
1720 The maximum send socket buffer size in bytes.
1722 message_burst and message_cost
1723 ------------------------------
1725 These parameters are used to limit the warning messages written to the kernel
1726 log from the networking code. They enforce a rate limit to make a
1727 denial-of-service attack impossible. A higher message_cost factor, results in
1728 fewer messages that will be written. Message_burst controls when messages will
1729 be dropped. The default settings limit warning messages to one every five
1735 This controls console messages from the networking stack that can occur because
1736 of problems on the network like duplicate address or bad checksums. Normally,
1737 this should be enabled, but if the problem persists the messages can be
1744 Maximum number of packets, queued on the INPUT side, when the interface
1745 receives packets faster than kernel can process them.
1750 Maximum ancillary buffer size allowed per socket. Ancillary data is a sequence
1751 of struct cmsghdr structures with appended data.
1753 /proc/sys/net/unix - Parameters for Unix domain sockets
1754 -------------------------------------------------------
1756 There are only two files in this subdirectory. They control the delays for
1757 deleting and destroying socket descriptors.
1759 2.8 /proc/sys/net/ipv4 - IPV4 settings
1760 --------------------------------------
1762 IP version 4 is still the most used protocol in Unix networking. It will be
1763 replaced by IP version 6 in the next couple of years, but for the moment it's
1764 the de facto standard for the internet and is used in most networking
1765 environments around the world. Because of the importance of this protocol,
1766 we'll have a deeper look into the subtree controlling the behavior of the IPv4
1767 subsystem of the Linux kernel.
1769 Let's start with the entries in /proc/sys/net/ipv4.
1774 icmp_echo_ignore_all and icmp_echo_ignore_broadcasts
1775 ----------------------------------------------------
1777 Turn on (1) or off (0), if the kernel should ignore all ICMP ECHO requests, or
1778 just those to broadcast and multicast addresses.
1780 Please note that if you accept ICMP echo requests with a broadcast/multi\-cast
1781 destination address your network may be used as an exploder for denial of
1782 service packet flooding attacks to other hosts.
1784 icmp_destunreach_rate, icmp_echoreply_rate, icmp_paramprob_rate and icmp_timeexeed_rate
1785 ---------------------------------------------------------------------------------------
1787 Sets limits for sending ICMP packets to specific targets. A value of zero
1788 disables all limiting. Any positive value sets the maximum package rate in
1789 hundredth of a second (on Intel systems).
1797 This file contains the number one if the host received its IP configuration by
1798 RARP, BOOTP, DHCP or a similar mechanism. Otherwise it is zero.
1803 TTL (Time To Live) for IPv4 interfaces. This is simply the maximum number of
1804 hops a packet may travel.
1809 Enable dynamic socket address rewriting on interface address change. This is
1810 useful for dialup interface with changing IP addresses.
1815 Enable or disable forwarding of IP packages between interfaces. Changing this
1816 value resets all other parameters to their default values. They differ if the
1817 kernel is configured as host or router.
1822 Range of ports used by TCP and UDP to choose the local port. Contains two
1823 numbers, the first number is the lowest port, the second number the highest
1824 local port. Default is 1024-4999. Should be changed to 32768-61000 for
1830 Global switch to turn path MTU discovery off. It can also be set on a per
1831 socket basis by the applications or on a per route basis.
1836 Enable/disable debugging of IP masquerading.
1838 IP fragmentation settings
1839 -------------------------
1841 ipfrag_high_trash and ipfrag_low_trash
1842 --------------------------------------
1844 Maximum memory used to reassemble IP fragments. When ipfrag_high_thresh bytes
1845 of memory is allocated for this purpose, the fragment handler will toss
1846 packets until ipfrag_low_thresh is reached.
1851 Time in seconds to keep an IP fragment in memory.
1859 This file controls the use of the ECN bit in the IPv4 headers. This is a new
1860 feature about Explicit Congestion Notification, but some routers and firewalls
1861 block traffic that has this bit set, so it could be necessary to echo 0 to
1862 /proc/sys/net/ipv4/tcp_ecn if you want to talk to these sites. For more info
1863 you could read RFC2481.
1865 tcp_retrans_collapse
1866 --------------------
1868 Bug-to-bug compatibility with some broken printers. On retransmit, try to send
1869 larger packets to work around bugs in certain TCP stacks. Can be turned off by
1872 tcp_keepalive_probes
1873 --------------------
1875 Number of keep alive probes TCP sends out, until it decides that the
1876 connection is broken.
1881 How often TCP sends out keep alive messages, when keep alive is enabled. The
1887 Number of times initial SYNs for a TCP connection attempt will be
1888 retransmitted. Should not be higher than 255. This is only the timeout for
1889 outgoing connections, for incoming connections the number of retransmits is
1890 defined by tcp_retries1.
1895 Enable select acknowledgments after RFC2018.
1900 Enable timestamps as defined in RFC1323.
1905 Enable the strict RFC793 interpretation of the TCP urgent pointer field. The
1906 default is to use the BSD compatible interpretation of the urgent pointer
1907 pointing to the first byte after the urgent data. The RFC793 interpretation is
1908 to have it point to the last byte of urgent data. Enabling this option may
1909 lead to interoperability problems. Disabled by default.
1914 Only valid when the kernel was compiled with CONFIG_SYNCOOKIES. Send out
1915 syncookies when the syn backlog queue of a socket overflows. This is to ward
1916 off the common 'syn flood attack'. Disabled by default.
1918 Note that the concept of a socket backlog is abandoned. This means the peer
1919 may not receive reliable error messages from an over loaded server with
1925 Enable window scaling as defined in RFC1323.
1930 The length of time in seconds it takes to receive a final FIN before the
1931 socket is always closed. This is strictly a violation of the TCP
1932 specification, but required to prevent denial-of-service attacks.
1937 Indicates how many keep alive probes are sent per slow timer run. Should not
1938 be set too high to prevent bursts.
1943 Length of the per socket backlog queue. Since Linux 2.2 the backlog specified
1944 in listen(2) only specifies the length of the backlog queue of already
1945 established sockets. When more connection requests arrive Linux starts to drop
1946 packets. When syncookies are enabled the packets are still answered and the
1947 maximum queue is effectively ignored.
1952 Defines how often an answer to a TCP connection request is retransmitted
1958 Defines how often a TCP packet is retransmitted before giving up.
1960 Interface specific settings
1961 ---------------------------
1963 In the directory /proc/sys/net/ipv4/conf you'll find one subdirectory for each
1964 interface the system knows about and one directory calls all. Changes in the
1965 all subdirectory affect all interfaces, whereas changes in the other
1966 subdirectories affect only one interface. All directories have the same
1972 This switch decides if the kernel accepts ICMP redirect messages or not. The
1973 default is 'yes' if the kernel is configured for a regular host and 'no' for a
1974 router configuration.
1979 Should source routed packages be accepted or declined. The default is
1980 dependent on the kernel configuration. It's 'yes' for routers and 'no' for
1986 Accept packets with source address 0.b.c.d with destinations not to this host
1987 as local ones. It is supposed that a BOOTP relay daemon will catch and forward
1990 The default is 0, since this feature is not implemented yet (kernel version
1996 Enable or disable IP forwarding on this interface.
2001 Log packets with source addresses with no known route to kernel log.
2006 Do multicast routing. The kernel needs to be compiled with CONFIG_MROUTE and a
2007 multicast routing daemon is required.
2012 Does (1) or does not (0) perform proxy ARP.
2017 Integer value determines if a source validation should be made. 1 means yes, 0
2018 means no. Disabled by default, but local/broadcast address spoofing is always
2021 If you set this to 1 on a router that is the only connection for a network to
2022 the net, it will prevent spoofing attacks against your internal networks
2023 (external addresses can still be spoofed), without the need for additional
2029 Accept ICMP redirect messages only for gateways, listed in default gateway
2030 list. Enabled by default.
2035 If it is not set the kernel does not assume that different subnets on this
2036 device can communicate directly. Default setting is 'yes'.
2041 Determines whether to send ICMP redirects to other hosts.
2046 The directory /proc/sys/net/ipv4/route contains several file to control
2049 error_burst and error_cost
2050 --------------------------
2052 These parameters are used to limit how many ICMP destination unreachable to
2053 send from the host in question. ICMP destination unreachable messages are
2054 sent when we cannot reach the next hop while trying to transmit a packet.
2055 It will also print some error messages to kernel logs if someone is ignoring
2056 our ICMP redirects. The higher the error_cost factor is, the fewer
2057 destination unreachable and error messages will be let through. Error_burst
2058 controls when destination unreachable messages and error messages will be
2059 dropped. The default settings limit warning messages to five every second.
2064 Writing to this file results in a flush of the routing cache.
2066 gc_elasticity, gc_interval, gc_min_interval_ms, gc_timeout, gc_thresh
2067 ---------------------------------------------------------------------
2069 Values to control the frequency and behavior of the garbage collection
2070 algorithm for the routing cache. gc_min_interval is deprecated and replaced
2071 by gc_min_interval_ms.
2077 Maximum size of the routing cache. Old entries will be purged once the cache
2078 reached has this size.
2080 redirect_load, redirect_number
2081 ------------------------------
2083 Factors which determine if more ICPM redirects should be sent to a specific
2084 host. No redirects will be sent once the load limit or the maximum number of
2085 redirects has been reached.
2090 Timeout for redirects. After this period redirects will be sent again, even if
2091 this has been stopped, because the load or number limit has been reached.
2093 Network Neighbor handling
2094 -------------------------
2096 Settings about how to handle connections with direct neighbors (nodes attached
2097 to the same link) can be found in the directory /proc/sys/net/ipv4/neigh.
2099 As we saw it in the conf directory, there is a default subdirectory which
2100 holds the default values, and one directory for each interface. The contents
2101 of the directories are identical, with the single exception that the default
2102 settings contain additional options to set garbage collection parameters.
2104 In the interface directories you'll find the following entries:
2106 base_reachable_time, base_reachable_time_ms
2107 -------------------------------------------
2109 A base value used for computing the random reachable time value as specified
2112 Expression of base_reachable_time, which is deprecated, is in seconds.
2113 Expression of base_reachable_time_ms is in milliseconds.
2115 retrans_time, retrans_time_ms
2116 -----------------------------
2118 The time between retransmitted Neighbor Solicitation messages.
2119 Used for address resolution and to determine if a neighbor is
2122 Expression of retrans_time, which is deprecated, is in 1/100 seconds (for
2123 IPv4) or in jiffies (for IPv6).
2124 Expression of retrans_time_ms is in milliseconds.
2129 Maximum queue length for a pending arp request - the number of packets which
2130 are accepted from other layers while the ARP address is still resolved.
2135 Maximum for random delay of answers to neighbor solicitation messages in
2136 jiffies (1/100 sec). Not yet implemented (Linux does not have anycast support
2142 Maximum number of retries for unicast solicitation.
2147 Maximum number of retries for multicast solicitation.
2149 delay_first_probe_time
2150 ----------------------
2152 Delay for the first time probe if the neighbor is reachable. (see
2158 An ARP/neighbor entry is only replaced with a new one if the old is at least
2159 locktime old. This prevents ARP cache thrashing.
2164 Maximum time (real time is random [0..proxytime]) before answering to an ARP
2165 request for which we have an proxy ARP entry. In some cases, this is used to
2166 prevent network flooding.
2171 Maximum queue length of the delayed proxy arp timer. (see proxy_delay).
2176 Determines the number of requests to send to the user level ARP daemon. Use 0
2182 Determines how often to check for stale ARP entries. After an ARP entry is
2183 stale it will be resolved again (which is useful when an IP address migrates
2184 to another machine). When ucast_solicit is greater than 0 it first tries to
2185 send an ARP packet directly to the known host When that fails and
2186 mcast_solicit is greater than 0, an ARP request is broadcasted.
2191 The /proc/sys/net/appletalk directory holds the Appletalk configuration data
2192 when Appletalk is loaded. The configurable parameters are:
2197 The amount of time we keep an ARP entry before expiring it. Used to age out
2203 The amount of time we will spend trying to resolve an Appletalk address.
2205 aarp-retransmit-limit
2206 ---------------------
2208 The number of times we will retransmit a query before giving up.
2213 Controls the rate at which expires are checked.
2215 The directory /proc/net/appletalk holds the list of active Appletalk sockets
2218 The fields indicate the DDP type, the local address (in network:node format)
2219 the remote address, the size of the transmit pending queue, the size of the
2220 received queue (bytes waiting for applications to read) the state and the uid
2223 /proc/net/atalk_iface lists all the interfaces configured for appletalk.It
2224 shows the name of the interface, its Appletalk address, the network range on
2225 that address (or network number for phase 1 networks), and the status of the
2228 /proc/net/atalk_route lists each known network route. It lists the target
2229 (network) that the route leads to, the router (may be directly connected), the
2230 route flags, and the device the route is using.
2235 The IPX protocol has no tunable values in proc/sys/net.
2237 The IPX protocol does, however, provide proc/net/ipx. This lists each IPX
2238 socket giving the local and remote addresses in Novell format (that is
2239 network:node:port). In accordance with the strange Novell tradition,
2240 everything but the port is in hex. Not_Connected is displayed for sockets that
2241 are not tied to a specific remote address. The Tx and Rx queue sizes indicate
2242 the number of bytes pending for transmission and reception. The state
2243 indicates the state the socket is in and the uid is the owning uid of the
2246 The /proc/net/ipx_interface file lists all IPX interfaces. For each interface
2247 it gives the network number, the node number, and indicates if the network is
2248 the primary network. It also indicates which device it is bound to (or
2249 Internal for internal networks) and the Frame Type if appropriate. Linux
2250 supports 802.3, 802.2, 802.2 SNAP and DIX (Blue Book) ethernet framing for
2253 The /proc/net/ipx_route table holds a list of IPX routes. For each route it
2254 gives the destination network, the router node (or Directly) and the network
2255 address of the router (or Connected) for internal networks.
2257 2.11 /proc/sys/fs/mqueue - POSIX message queues filesystem
2258 ----------------------------------------------------------
2260 The "mqueue" filesystem provides the necessary kernel features to enable the
2261 creation of a user space library that implements the POSIX message queues
2262 API (as noted by the MSG tag in the POSIX 1003.1-2001 version of the System
2263 Interfaces specification.)
2265 The "mqueue" filesystem contains values for determining/setting the amount of
2266 resources used by the file system.
2268 /proc/sys/fs/mqueue/queues_max is a read/write file for setting/getting the
2269 maximum number of message queues allowed on the system.
2271 /proc/sys/fs/mqueue/msg_max is a read/write file for setting/getting the
2272 maximum number of messages in a queue value. In fact it is the limiting value
2273 for another (user) limit which is set in mq_open invocation. This attribute of
2274 a queue must be less or equal then msg_max.
2276 /proc/sys/fs/mqueue/msgsize_max is a read/write file for setting/getting the
2277 maximum message size value (it is every message queue's attribute set during
2280 2.12 /proc/<pid>/oom_adj - Adjust the oom-killer score
2281 ------------------------------------------------------
2283 This file can be used to adjust the score used to select which processes
2284 should be killed in an out-of-memory situation. Giving it a high score will
2285 increase the likelihood of this process being killed by the oom-killer. Valid
2286 values are in the range -16 to +15, plus the special value -17, which disables
2287 oom-killing altogether for this process.
2289 2.13 /proc/<pid>/oom_score - Display current oom-killer score
2290 -------------------------------------------------------------
2292 ------------------------------------------------------------------------------
2293 This file can be used to check the current score used by the oom-killer is for
2294 any given <pid>. Use it together with /proc/<pid>/oom_adj to tune which
2295 process should be killed in an out-of-memory situation.
2297 ------------------------------------------------------------------------------
2299 ------------------------------------------------------------------------------
2300 Certain aspects of kernel behavior can be modified at runtime, without the
2301 need to recompile the kernel, or even to reboot the system. The files in the
2302 /proc/sys tree can not only be read, but also modified. You can use the echo
2303 command to write value into these files, thereby changing the default settings
2305 ------------------------------------------------------------------------------
2307 2.14 /proc/<pid>/io - Display the IO accounting fields
2308 -------------------------------------------------------
2310 This file contains IO statistics for each running process
2315 test:/tmp # dd if=/dev/zero of=/tmp/test.dat &
2318 test:/tmp # cat /proc/3828/io
2324 write_bytes: 323932160
2325 cancelled_write_bytes: 0
2334 I/O counter: chars read
2335 The number of bytes which this task has caused to be read from storage. This
2336 is simply the sum of bytes which this process passed to read() and pread().
2337 It includes things like tty IO and it is unaffected by whether or not actual
2338 physical disk IO was required (the read might have been satisfied from
2345 I/O counter: chars written
2346 The number of bytes which this task has caused, or shall cause to be written
2347 to disk. Similar caveats apply here as with rchar.
2353 I/O counter: read syscalls
2354 Attempt to count the number of read I/O operations, i.e. syscalls like read()
2361 I/O counter: write syscalls
2362 Attempt to count the number of write I/O operations, i.e. syscalls like
2363 write() and pwrite().
2369 I/O counter: bytes read
2370 Attempt to count the number of bytes which this process really did cause to
2371 be fetched from the storage layer. Done at the submit_bio() level, so it is
2372 accurate for block-backed filesystems. <please add status regarding NFS and
2373 CIFS at a later time>
2379 I/O counter: bytes written
2380 Attempt to count the number of bytes which this process caused to be sent to
2381 the storage layer. This is done at page-dirtying time.
2384 cancelled_write_bytes
2385 ---------------------
2387 The big inaccuracy here is truncate. If a process writes 1MB to a file and
2388 then deletes the file, it will in fact perform no writeout. But it will have
2389 been accounted as having caused 1MB of write.
2390 In other words: The number of bytes which this process caused to not happen,
2391 by truncating pagecache. A task can cause "negative" IO too. If this task
2392 truncates some dirty pagecache, some IO which another task has been accounted
2393 for (in it's write_bytes) will not be happening. We _could_ just subtract that
2394 from the truncating task's write_bytes, but there is information loss in doing
2401 At its current implementation state, this is a bit racy on 32-bit machines: if
2402 process A reads process B's /proc/pid/io while process B is updating one of
2403 those 64-bit counters, process A could see an intermediate result.
2406 More information about this can be found within the taskstats documentation in
2407 Documentation/accounting.
2409 2.15 /proc/<pid>/coredump_filter - Core dump filtering settings
2410 ---------------------------------------------------------------
2411 When a process is dumped, all anonymous memory is written to a core file as
2412 long as the size of the core file isn't limited. But sometimes we don't want
2413 to dump some memory segments, for example, huge shared memory. Conversely,
2414 sometimes we want to save file-backed memory segments into a core file, not
2415 only the individual files.
2417 /proc/<pid>/coredump_filter allows you to customize which memory segments
2418 will be dumped when the <pid> process is dumped. coredump_filter is a bitmask
2419 of memory types. If a bit of the bitmask is set, memory segments of the
2420 corresponding memory type are dumped, otherwise they are not dumped.
2422 The following 7 memory types are supported:
2423 - (bit 0) anonymous private memory
2424 - (bit 1) anonymous shared memory
2425 - (bit 2) file-backed private memory
2426 - (bit 3) file-backed shared memory
2427 - (bit 4) ELF header pages in file-backed private memory areas (it is
2428 effective only if the bit 2 is cleared)
2429 - (bit 5) hugetlb private memory
2430 - (bit 6) hugetlb shared memory
2432 Note that MMIO pages such as frame buffer are never dumped and vDSO pages
2433 are always dumped regardless of the bitmask status.
2435 Note bit 0-4 doesn't effect any hugetlb memory. hugetlb memory are only
2436 effected by bit 5-6.
2438 Default value of coredump_filter is 0x23; this means all anonymous memory
2439 segments and hugetlb private memory are dumped.
2441 If you don't want to dump all shared memory segments attached to pid 1234,
2442 write 0x21 to the process's proc file.
2444 $ echo 0x21 > /proc/1234/coredump_filter
2446 When a new process is created, the process inherits the bitmask status from its
2447 parent. It is useful to set up coredump_filter before the program runs.
2450 $ echo 0x7 > /proc/self/coredump_filter
2453 2.16 /proc/<pid>/mountinfo - Information about mounts
2454 --------------------------------------------------------
2456 This file contains lines of the form:
2458 36 35 98:0 /mnt1 /mnt2 rw,noatime master:1 - ext3 /dev/root rw,errors=continue
2459 (1)(2)(3) (4) (5) (6) (7) (8) (9) (10) (11)
2461 (1) mount ID: unique identifier of the mount (may be reused after umount)
2462 (2) parent ID: ID of parent (or of self for the top of the mount tree)
2463 (3) major:minor: value of st_dev for files on filesystem
2464 (4) root: root of the mount within the filesystem
2465 (5) mount point: mount point relative to the process's root
2466 (6) mount options: per mount options
2467 (7) optional fields: zero or more fields of the form "tag[:value]"
2468 (8) separator: marks the end of the optional fields
2469 (9) filesystem type: name of filesystem of the form "type[.subtype]"
2470 (10) mount source: filesystem specific information or "none"
2471 (11) super options: per super block options
2473 Parsers should ignore all unrecognised optional fields. Currently the
2474 possible optional fields are:
2476 shared:X mount is shared in peer group X
2477 master:X mount is slave to peer group X
2478 propagate_from:X mount is slave and receives propagation from peer group X (*)
2479 unbindable mount is unbindable
2481 (*) X is the closest dominant peer group under the process's root. If
2482 X is the immediate master of the mount, or if there's no dominant peer
2483 group under the same root, then only the "master:X" field is present
2484 and not the "propagate_from:X" field.
2486 For more information on mount propagation see:
2488 Documentation/filesystems/sharedsubtree.txt
2490 2.17 /proc/sys/fs/epoll - Configuration options for the epoll interface
2491 --------------------------------------------------------
2493 This directory contains configuration options for the epoll(7) interface.
2498 This is the maximum number of epoll file descriptors that a single user can
2499 have open at a given time. The default value is 128, and should be enough
2505 Every epoll file descriptor can store a number of files to be monitored
2506 for event readiness. Each one of these monitored files constitutes a "watch".
2507 This configuration option sets the maximum number of "watches" that are
2508 allowed for each user.
2509 Each "watch" costs roughly 90 bytes on a 32bit kernel, and roughly 160 bytes
2511 The current default value for max_user_watches is the 1/32 of the available
2512 low memory, divided for the "watch" cost in bytes.
2515 ------------------------------------------------------------------------------