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
47 ------------------------------------------------------------------------------
49 ------------------------------------------------------------------------------
51 0.1 Introduction/Credits
52 ------------------------
54 This documentation is part of a soon (or so we hope) to be released book on
55 the SuSE Linux distribution. As there is no complete documentation for the
56 /proc file system and we've used many freely available sources to write these
57 chapters, it seems only fair to give the work back to the Linux community.
58 This work is based on the 2.2.* kernel version and the upcoming 2.4.*. I'm
59 afraid it's still far from complete, but we hope it will be useful. As far as
60 we know, it is the first 'all-in-one' document about the /proc file system. It
61 is focused on the Intel x86 hardware, so if you are looking for PPC, ARM,
62 SPARC, AXP, etc., features, you probably won't find what you are looking for.
63 It also only covers IPv4 networking, not IPv6 nor other protocols - sorry. But
64 additions and patches are welcome and will be added to this document if you
67 We'd like to thank Alan Cox, Rik van Riel, and Alexey Kuznetsov and a lot of
68 other people for help compiling this documentation. We'd also like to extend a
69 special thank you to Andi Kleen for documentation, which we relied on heavily
70 to create this document, as well as the additional information he provided.
71 Thanks to everybody else who contributed source or docs to the Linux kernel
72 and helped create a great piece of software... :)
74 If you have any comments, corrections or additions, please don't hesitate to
75 contact Bodo Bauer at bb@ricochet.net. We'll be happy to add them to this
78 The latest version of this document is available online at
79 http://skaro.nightcrawler.com/~bb/Docs/Proc as HTML version.
81 If the above direction does not works for you, ypu could try the kernel
82 mailing list at linux-kernel@vger.kernel.org and/or try to reach me at
83 comandante@zaralinux.com.
88 We don't guarantee the correctness of this document, and if you come to us
89 complaining about how you screwed up your system because of incorrect
90 documentation, we won't feel responsible...
92 ------------------------------------------------------------------------------
93 CHAPTER 1: COLLECTING SYSTEM INFORMATION
94 ------------------------------------------------------------------------------
96 ------------------------------------------------------------------------------
98 ------------------------------------------------------------------------------
99 * Investigating the properties of the pseudo file system /proc and its
100 ability to provide information on the running Linux system
101 * Examining /proc's structure
102 * Uncovering various information about the kernel and the processes running
104 ------------------------------------------------------------------------------
107 The proc file system acts as an interface to internal data structures in the
108 kernel. It can be used to obtain information about the system and to change
109 certain kernel parameters at runtime (sysctl).
111 First, we'll take a look at the read-only parts of /proc. In Chapter 2, we
112 show you how you can use /proc/sys to change settings.
114 1.1 Process-Specific Subdirectories
115 -----------------------------------
117 The directory /proc contains (among other things) one subdirectory for each
118 process running on the system, which is named after the process ID (PID).
120 The link self points to the process reading the file system. Each process
121 subdirectory has the entries listed in Table 1-1.
124 Table 1-1: Process specific entries in /proc
125 ..............................................................................
127 clear_refs Clears page referenced bits shown in smaps output
128 cmdline Command line arguments
129 cpu Current and last cpu in which it was executed (2.4)(smp)
130 cwd Link to the current working directory
131 environ Values of environment variables
132 exe Link to the executable of this process
133 fd Directory, which contains all file descriptors
134 maps Memory maps to executables and library files (2.4)
135 mem Memory held by this process
136 root Link to the root directory of this process
138 statm Process memory status information
139 status Process status in human readable form
140 wchan If CONFIG_KALLSYMS is set, a pre-decoded wchan
141 smaps Extension based on maps, the rss size for each mapped file
142 ..............................................................................
144 For example, to get the status information of a process, all you have to do is
145 read the file /proc/PID/status:
147 >cat /proc/self/status
163 SigPnd: 0000000000000000
164 SigBlk: 0000000000000000
165 SigIgn: 0000000000000000
166 SigCgt: 0000000000000000
167 CapInh: 00000000fffffeff
168 CapPrm: 0000000000000000
169 CapEff: 0000000000000000
172 This shows you nearly the same information you would get if you viewed it with
173 the ps command. In fact, ps uses the proc file system to obtain its
174 information. The statm file contains more detailed information about the
175 process memory usage. Its seven fields are explained in Table 1-2. The stat
176 file contains details information about the process itself. Its fields are
177 explained in Table 1-3.
180 Table 1-2: Contents of the statm files (as of 2.6.8-rc3)
181 ..............................................................................
183 size total program size (pages) (same as VmSize in status)
184 resident size of memory portions (pages) (same as VmRSS in status)
185 shared number of pages that are shared (i.e. backed by a file)
186 trs number of pages that are 'code' (not including libs; broken,
187 includes data segment)
188 lrs number of pages of library (always 0 on 2.6)
189 drs number of pages of data/stack (including libs; broken,
190 includes library text)
191 dt number of dirty pages (always 0 on 2.6)
192 ..............................................................................
195 Table 1-3: Contents of the stat files (as of 2.6.22-rc3)
196 ..............................................................................
199 tcomm filename of the executable
200 state state (R is running, S is sleeping, D is sleeping in an
201 uninterruptible wait, Z is zombie, T is traced or stopped)
202 ppid process id of the parent process
203 pgrp pgrp of the process
205 tty_nr tty the process uses
206 tty_pgrp pgrp of the tty
208 min_flt number of minor faults
209 cmin_flt number of minor faults with child's
210 maj_flt number of major faults
211 cmaj_flt number of major faults with child's
212 utime user mode jiffies
213 stime kernel mode jiffies
214 cutime user mode jiffies with child's
215 cstime kernel mode jiffies with child's
216 priority priority level
218 num_threads number of threads
219 it_real_value (obsolete, always 0)
220 start_time time the process started after system boot
221 vsize virtual memory size
222 rss resident set memory size
223 rsslim current limit in bytes on the rss
224 start_code address above which program text can run
225 end_code address below which program text can run
226 start_stack address of the start of the stack
227 esp current value of ESP
228 eip current value of EIP
229 pending bitmap of pending signals (obsolete)
230 blocked bitmap of blocked signals (obsolete)
231 sigign bitmap of ignored signals (obsolete)
232 sigcatch bitmap of catched signals (obsolete)
233 wchan address where process went to sleep
236 exit_signal signal to send to parent thread on exit
237 task_cpu which CPU the task is scheduled on
238 rt_priority realtime priority
239 policy scheduling policy (man sched_setscheduler)
240 blkio_ticks time spent waiting for block IO
241 ..............................................................................
247 Similar to the process entries, the kernel data files give information about
248 the running kernel. The files used to obtain this information are contained in
249 /proc and are listed in Table 1-4. Not all of these will be present in your
250 system. It depends on the kernel configuration and the loaded modules, which
251 files are there, and which are missing.
253 Table 1-4: Kernel info in /proc
254 ..............................................................................
256 apm Advanced power management info
257 buddyinfo Kernel memory allocator information (see text) (2.5)
258 bus Directory containing bus specific information
259 cmdline Kernel command line
260 cpuinfo Info about the CPU
261 devices Available devices (block and character)
262 dma Used DMS channels
263 filesystems Supported filesystems
264 driver Various drivers grouped here, currently rtc (2.4)
265 execdomains Execdomains, related to security (2.4)
266 fb Frame Buffer devices (2.4)
267 fs File system parameters, currently nfs/exports (2.4)
268 ide Directory containing info about the IDE subsystem
269 interrupts Interrupt usage
270 iomem Memory map (2.4)
271 ioports I/O port usage
272 irq Masks for irq to cpu affinity (2.4)(smp?)
273 isapnp ISA PnP (Plug&Play) Info (2.4)
274 kcore Kernel core image (can be ELF or A.OUT(deprecated in 2.4))
276 ksyms Kernel symbol table
277 loadavg Load average of last 1, 5 & 15 minutes
281 modules List of loaded modules
282 mounts Mounted filesystems
283 net Networking info (see text)
284 partitions Table of partitions known to the system
285 pci Deprecated info of PCI bus (new way -> /proc/bus/pci/,
286 decoupled by lspci (2.4)
288 scsi SCSI info (see text)
289 slabinfo Slab pool info
290 stat Overall statistics
291 swaps Swap space utilization
293 sysvipc Info of SysVIPC Resources (msg, sem, shm) (2.4)
294 tty Info of tty drivers
296 version Kernel version
297 video bttv info of video resources (2.4)
298 ..............................................................................
300 You can, for example, check which interrupts are currently in use and what
301 they are used for by looking in the file /proc/interrupts:
303 > cat /proc/interrupts
305 0: 8728810 XT-PIC timer
306 1: 895 XT-PIC keyboard
308 3: 531695 XT-PIC aha152x
309 4: 2014133 XT-PIC serial
310 5: 44401 XT-PIC pcnet_cs
313 12: 182918 XT-PIC PS/2 Mouse
315 14: 1232265 XT-PIC ide0
319 In 2.4.* a couple of lines where added to this file LOC & ERR (this time is the
320 output of a SMP machine):
322 > cat /proc/interrupts
325 0: 1243498 1214548 IO-APIC-edge timer
326 1: 8949 8958 IO-APIC-edge keyboard
327 2: 0 0 XT-PIC cascade
328 5: 11286 10161 IO-APIC-edge soundblaster
329 8: 1 0 IO-APIC-edge rtc
330 9: 27422 27407 IO-APIC-edge 3c503
331 12: 113645 113873 IO-APIC-edge PS/2 Mouse
333 14: 22491 24012 IO-APIC-edge ide0
334 15: 2183 2415 IO-APIC-edge ide1
335 17: 30564 30414 IO-APIC-level eth0
336 18: 177 164 IO-APIC-level bttv
341 NMI is incremented in this case because every timer interrupt generates a NMI
342 (Non Maskable Interrupt) which is used by the NMI Watchdog to detect lockups.
344 LOC is the local interrupt counter of the internal APIC of every CPU.
346 ERR is incremented in the case of errors in the IO-APIC bus (the bus that
347 connects the CPUs in a SMP system. This means that an error has been detected,
348 the IO-APIC automatically retry the transmission, so it should not be a big
349 problem, but you should read the SMP-FAQ.
351 In 2.6.2* /proc/interrupts was expanded again. This time the goal was for
352 /proc/interrupts to display every IRQ vector in use by the system, not
353 just those considered 'most important'. The new vectors are:
355 THR -- interrupt raised when a machine check threshold counter
356 (typically counting ECC corrected errors of memory or cache) exceeds
357 a configurable threshold. Only available on some systems.
359 TRM -- a thermal event interrupt occurs when a temperature threshold
360 has been exceeded for the CPU. This interrupt may also be generated
361 when the temperature drops back to normal.
363 SPU -- a spurious interrupt is some interrupt that was raised then lowered
364 by some IO device before it could be fully processed by the APIC. Hence
365 the APIC sees the interrupt but does not know what device it came from.
366 For this case the APIC will generate the interrupt with a IRQ vector
367 of 0xff. This might also be generated by chipset bugs.
369 RES, CAL, TLB -- rescheduling, call and TLB flush interrupts are
370 sent from one CPU to another per the needs of the OS. Typically,
371 their statistics are used by kernel developers and interested users to
372 determine the occurance of interrupt of the given type.
374 The above IRQ vectors are displayed only when relevent. For example,
375 the threshold vector does not exist on x86_64 platforms. Others are
376 suppressed when the system is a uniprocessor. As of this writing, only
377 i386 and x86_64 platforms support the new IRQ vector displays.
379 Of some interest is the introduction of the /proc/irq directory to 2.4.
380 It could be used to set IRQ to CPU affinity, this means that you can "hook" an
381 IRQ to only one CPU, or to exclude a CPU of handling IRQs. The contents of the
382 irq subdir is one subdir for each IRQ, and one file; prof_cpu_mask
386 0 10 12 14 16 18 2 4 6 8 prof_cpu_mask
387 1 11 13 15 17 19 3 5 7 9
391 The contents of the prof_cpu_mask file and each smp_affinity file for each IRQ
392 is the same by default:
394 > cat /proc/irq/0/smp_affinity
397 It's a bitmask, in which you can specify which CPUs can handle the IRQ, you can
400 > echo 1 > /proc/irq/prof_cpu_mask
402 This means that only the first CPU will handle the IRQ, but you can also echo 5
403 which means that only the first and fourth CPU can handle the IRQ.
405 The way IRQs are routed is handled by the IO-APIC, and it's Round Robin
406 between all the CPUs which are allowed to handle it. As usual the kernel has
407 more info than you and does a better job than you, so the defaults are the
408 best choice for almost everyone.
410 There are three more important subdirectories in /proc: net, scsi, and sys.
411 The general rule is that the contents, or even the existence of these
412 directories, depend on your kernel configuration. If SCSI is not enabled, the
413 directory scsi may not exist. The same is true with the net, which is there
414 only when networking support is present in the running kernel.
416 The slabinfo file gives information about memory usage at the slab level.
417 Linux uses slab pools for memory management above page level in version 2.2.
418 Commonly used objects have their own slab pool (such as network buffers,
419 directory cache, and so on).
421 ..............................................................................
423 > cat /proc/buddyinfo
425 Node 0, zone DMA 0 4 5 4 4 3 ...
426 Node 0, zone Normal 1 0 0 1 101 8 ...
427 Node 0, zone HighMem 2 0 0 1 1 0 ...
429 Memory fragmentation is a problem under some workloads, and buddyinfo is a
430 useful tool for helping diagnose these problems. Buddyinfo will give you a
431 clue as to how big an area you can safely allocate, or why a previous
434 Each column represents the number of pages of a certain order which are
435 available. In this case, there are 0 chunks of 2^0*PAGE_SIZE available in
436 ZONE_DMA, 4 chunks of 2^1*PAGE_SIZE in ZONE_DMA, 101 chunks of 2^4*PAGE_SIZE
437 available in ZONE_NORMAL, etc...
439 ..............................................................................
443 Provides information about distribution and utilization of memory. This
444 varies by architecture and compile options. The following is from a
445 16GB PIII, which has highmem enabled. You may not have all of these fields.
450 MemTotal: 16344972 kB
457 HighTotal: 15597528 kB
458 HighFree: 13629632 kB
467 CommitLimit: 7669796 kB
468 Committed_AS: 100056 kB
470 VmallocTotal: 112216 kB
472 VmallocChunk: 111088 kB
474 MemTotal: Total usable ram (i.e. physical ram minus a few reserved
475 bits and the kernel binary code)
476 MemFree: The sum of LowFree+HighFree
477 Buffers: Relatively temporary storage for raw disk blocks
478 shouldn't get tremendously large (20MB or so)
479 Cached: in-memory cache for files read from the disk (the
480 pagecache). Doesn't include SwapCached
481 SwapCached: Memory that once was swapped out, is swapped back in but
482 still also is in the swapfile (if memory is needed it
483 doesn't need to be swapped out AGAIN because it is already
484 in the swapfile. This saves I/O)
485 Active: Memory that has been used more recently and usually not
486 reclaimed unless absolutely necessary.
487 Inactive: Memory which has been less recently used. It is more
488 eligible to be reclaimed for other purposes
490 HighFree: Highmem is all memory above ~860MB of physical memory
491 Highmem areas are for use by userspace programs, or
492 for the pagecache. The kernel must use tricks to access
493 this memory, making it slower to access than lowmem.
495 LowFree: Lowmem is memory which can be used for everything that
496 highmem can be used for, but it is also available for the
497 kernel's use for its own data structures. Among many
498 other things, it is where everything from the Slab is
499 allocated. Bad things happen when you're out of lowmem.
500 SwapTotal: total amount of swap space available
501 SwapFree: Memory which has been evicted from RAM, and is temporarily
503 Dirty: Memory which is waiting to get written back to the disk
504 Writeback: Memory which is actively being written back to the disk
505 Mapped: files which have been mmaped, such as libraries
506 Slab: in-kernel data structures cache
507 CommitLimit: Based on the overcommit ratio ('vm.overcommit_ratio'),
508 this is the total amount of memory currently available to
509 be allocated on the system. This limit is only adhered to
510 if strict overcommit accounting is enabled (mode 2 in
511 'vm.overcommit_memory').
512 The CommitLimit is calculated with the following formula:
513 CommitLimit = ('vm.overcommit_ratio' * Physical RAM) + Swap
514 For example, on a system with 1G of physical RAM and 7G
515 of swap with a `vm.overcommit_ratio` of 30 it would
516 yield a CommitLimit of 7.3G.
517 For more details, see the memory overcommit documentation
518 in vm/overcommit-accounting.
519 Committed_AS: The amount of memory presently allocated on the system.
520 The committed memory is a sum of all of the memory which
521 has been allocated by processes, even if it has not been
522 "used" by them as of yet. A process which malloc()'s 1G
523 of memory, but only touches 300M of it will only show up
524 as using 300M of memory even if it has the address space
525 allocated for the entire 1G. This 1G is memory which has
526 been "committed" to by the VM and can be used at any time
527 by the allocating application. With strict overcommit
528 enabled on the system (mode 2 in 'vm.overcommit_memory'),
529 allocations which would exceed the CommitLimit (detailed
530 above) will not be permitted. This is useful if one needs
531 to guarantee that processes will not fail due to lack of
532 memory once that memory has been successfully allocated.
533 PageTables: amount of memory dedicated to the lowest level of page
535 VmallocTotal: total size of vmalloc memory area
536 VmallocUsed: amount of vmalloc area which is used
537 VmallocChunk: largest contigious block of vmalloc area which is free
540 1.3 IDE devices in /proc/ide
541 ----------------------------
543 The subdirectory /proc/ide contains information about all IDE devices of which
544 the kernel is aware. There is one subdirectory for each IDE controller, the
545 file drivers and a link for each IDE device, pointing to the device directory
546 in the controller specific subtree.
548 The file drivers contains general information about the drivers used for the
551 > cat /proc/ide/drivers
552 ide-cdrom version 4.53
553 ide-disk version 1.08
555 More detailed information can be found in the controller specific
556 subdirectories. These are named ide0, ide1 and so on. Each of these
557 directories contains the files shown in table 1-5.
560 Table 1-5: IDE controller info in /proc/ide/ide?
561 ..............................................................................
563 channel IDE channel (0 or 1)
564 config Configuration (only for PCI/IDE bridge)
566 model Type/Chipset of IDE controller
567 ..............................................................................
569 Each device connected to a controller has a separate subdirectory in the
570 controllers directory. The files listed in table 1-6 are contained in these
574 Table 1-6: IDE device information
575 ..............................................................................
578 capacity Capacity of the medium (in 512Byte blocks)
579 driver driver and version
580 geometry physical and logical geometry
581 identify device identify block
583 model device identifier
584 settings device setup
585 smart_thresholds IDE disk management thresholds
586 smart_values IDE disk management values
587 ..............................................................................
589 The most interesting file is settings. This file contains a nice overview of
590 the drive parameters:
592 # cat /proc/ide/ide0/hda/settings
593 name value min max mode
594 ---- ----- --- --- ----
595 bios_cyl 526 0 65535 rw
596 bios_head 255 0 255 rw
598 breada_readahead 4 0 127 rw
600 file_readahead 72 0 2097151 rw
602 keepsettings 0 0 1 rw
603 max_kb_per_request 122 1 127 rw
607 pio_mode write-only 0 255 w
613 1.4 Networking info in /proc/net
614 --------------------------------
616 The subdirectory /proc/net follows the usual pattern. Table 1-6 shows the
617 additional values you get for IP version 6 if you configure the kernel to
618 support this. Table 1-7 lists the files and their meaning.
621 Table 1-6: IPv6 info in /proc/net
622 ..............................................................................
624 udp6 UDP sockets (IPv6)
625 tcp6 TCP sockets (IPv6)
626 raw6 Raw device statistics (IPv6)
627 igmp6 IP multicast addresses, which this host joined (IPv6)
628 if_inet6 List of IPv6 interface addresses
629 ipv6_route Kernel routing table for IPv6
630 rt6_stats Global IPv6 routing tables statistics
631 sockstat6 Socket statistics (IPv6)
632 snmp6 Snmp data (IPv6)
633 ..............................................................................
636 Table 1-7: Network info in /proc/net
637 ..............................................................................
640 dev network devices with statistics
641 dev_mcast the Layer2 multicast groups a device is listening too
642 (interface index, label, number of references, number of bound
644 dev_stat network device status
645 ip_fwchains Firewall chain linkage
646 ip_fwnames Firewall chain names
647 ip_masq Directory containing the masquerading tables
648 ip_masquerade Major masquerading table
649 netstat Network statistics
650 raw raw device statistics
651 route Kernel routing table
652 rpc Directory containing rpc info
653 rt_cache Routing cache
655 sockstat Socket statistics
657 tr_rif Token ring RIF routing table
659 unix UNIX domain sockets
660 wireless Wireless interface data (Wavelan etc)
661 igmp IP multicast addresses, which this host joined
662 psched Global packet scheduler parameters.
663 netlink List of PF_NETLINK sockets
664 ip_mr_vifs List of multicast virtual interfaces
665 ip_mr_cache List of multicast routing cache
666 ..............................................................................
668 You can use this information to see which network devices are available in
669 your system and how much traffic was routed over those devices:
673 face |bytes packets errs drop fifo frame compressed multicast|[...
674 lo: 908188 5596 0 0 0 0 0 0 [...
675 ppp0:15475140 20721 410 0 0 410 0 0 [...
676 eth0: 614530 7085 0 0 0 0 0 1 [...
679 ...] bytes packets errs drop fifo colls carrier compressed
680 ...] 908188 5596 0 0 0 0 0 0
681 ...] 1375103 17405 0 0 0 0 0 0
682 ...] 1703981 5535 0 0 0 3 0 0
684 In addition, each Channel Bond interface has it's own directory. For
685 example, the bond0 device will have a directory called /proc/net/bond0/.
686 It will contain information that is specific to that bond, such as the
687 current slaves of the bond, the link status of the slaves, and how
688 many times the slaves link has failed.
693 If you have a SCSI host adapter in your system, you'll find a subdirectory
694 named after the driver for this adapter in /proc/scsi. You'll also see a list
695 of all recognized SCSI devices in /proc/scsi:
699 Host: scsi0 Channel: 00 Id: 00 Lun: 00
700 Vendor: IBM Model: DGHS09U Rev: 03E0
701 Type: Direct-Access ANSI SCSI revision: 03
702 Host: scsi0 Channel: 00 Id: 06 Lun: 00
703 Vendor: PIONEER Model: CD-ROM DR-U06S Rev: 1.04
704 Type: CD-ROM ANSI SCSI revision: 02
707 The directory named after the driver has one file for each adapter found in
708 the system. These files contain information about the controller, including
709 the used IRQ and the IO address range. The amount of information shown is
710 dependent on the adapter you use. The example shows the output for an Adaptec
711 AHA-2940 SCSI adapter:
713 > cat /proc/scsi/aic7xxx/0
715 Adaptec AIC7xxx driver version: 5.1.19/3.2.4
717 TCQ Enabled By Default : Disabled
718 AIC7XXX_PROC_STATS : Disabled
719 AIC7XXX_RESET_DELAY : 5
720 Adapter Configuration:
721 SCSI Adapter: Adaptec AHA-294X Ultra SCSI host adapter
722 Ultra Wide Controller
723 PCI MMAPed I/O Base: 0xeb001000
724 Adapter SEEPROM Config: SEEPROM found and used.
725 Adaptec SCSI BIOS: Enabled
727 SCBs: Active 0, Max Active 2,
728 Allocated 15, HW 16, Page 255
730 BIOS Control Word: 0x18b6
731 Adapter Control Word: 0x005b
732 Extended Translation: Enabled
733 Disconnect Enable Flags: 0xffff
734 Ultra Enable Flags: 0x0001
735 Tag Queue Enable Flags: 0x0000
736 Ordered Queue Tag Flags: 0x0000
737 Default Tag Queue Depth: 8
738 Tagged Queue By Device array for aic7xxx host instance 0:
739 {255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255}
740 Actual queue depth per device for aic7xxx host instance 0:
741 {1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1}
744 Device using Wide/Sync transfers at 40.0 MByte/sec, offset 8
745 Transinfo settings: current(12/8/1/0), goal(12/8/1/0), user(12/15/1/0)
746 Total transfers 160151 (74577 reads and 85574 writes)
748 Device using Narrow/Sync transfers at 5.0 MByte/sec, offset 15
749 Transinfo settings: current(50/15/0/0), goal(50/15/0/0), user(50/15/0/0)
750 Total transfers 0 (0 reads and 0 writes)
753 1.6 Parallel port info in /proc/parport
754 ---------------------------------------
756 The directory /proc/parport contains information about the parallel ports of
757 your system. It has one subdirectory for each port, named after the port
760 These directories contain the four files shown in Table 1-8.
763 Table 1-8: Files in /proc/parport
764 ..............................................................................
766 autoprobe Any IEEE-1284 device ID information that has been acquired.
767 devices list of the device drivers using that port. A + will appear by the
768 name of the device currently using the port (it might not appear
770 hardware Parallel port's base address, IRQ line and DMA channel.
771 irq IRQ that parport is using for that port. This is in a separate
772 file to allow you to alter it by writing a new value in (IRQ
774 ..............................................................................
776 1.7 TTY info in /proc/tty
777 -------------------------
779 Information about the available and actually used tty's can be found in the
780 directory /proc/tty.You'll find entries for drivers and line disciplines in
781 this directory, as shown in Table 1-9.
784 Table 1-9: Files in /proc/tty
785 ..............................................................................
787 drivers list of drivers and their usage
788 ldiscs registered line disciplines
789 driver/serial usage statistic and status of single tty lines
790 ..............................................................................
792 To see which tty's are currently in use, you can simply look into the file
795 > cat /proc/tty/drivers
796 pty_slave /dev/pts 136 0-255 pty:slave
797 pty_master /dev/ptm 128 0-255 pty:master
798 pty_slave /dev/ttyp 3 0-255 pty:slave
799 pty_master /dev/pty 2 0-255 pty:master
800 serial /dev/cua 5 64-67 serial:callout
801 serial /dev/ttyS 4 64-67 serial
802 /dev/tty0 /dev/tty0 4 0 system:vtmaster
803 /dev/ptmx /dev/ptmx 5 2 system
804 /dev/console /dev/console 5 1 system:console
805 /dev/tty /dev/tty 5 0 system:/dev/tty
806 unknown /dev/tty 4 1-63 console
809 1.8 Miscellaneous kernel statistics in /proc/stat
810 -------------------------------------------------
812 Various pieces of information about kernel activity are available in the
813 /proc/stat file. All of the numbers reported in this file are aggregates
814 since the system first booted. For a quick look, simply cat the file:
817 cpu 2255 34 2290 22625563 6290 127 456 0
818 cpu0 1132 34 1441 11311718 3675 127 438 0
819 cpu1 1123 0 849 11313845 2614 0 18 0
820 intr 114930548 113199788 3 0 5 263 0 4 [... lots more numbers ...]
827 The very first "cpu" line aggregates the numbers in all of the other "cpuN"
828 lines. These numbers identify the amount of time the CPU has spent performing
829 different kinds of work. Time units are in USER_HZ (typically hundredths of a
830 second). The meanings of the columns are as follows, from left to right:
832 - user: normal processes executing in user mode
833 - nice: niced processes executing in user mode
834 - system: processes executing in kernel mode
835 - idle: twiddling thumbs
836 - iowait: waiting for I/O to complete
837 - irq: servicing interrupts
838 - softirq: servicing softirqs
839 - steal: involuntary wait
841 The "intr" line gives counts of interrupts serviced since boot time, for each
842 of the possible system interrupts. The first column is the total of all
843 interrupts serviced; each subsequent column is the total for that particular
846 The "ctxt" line gives the total number of context switches across all CPUs.
848 The "btime" line gives the time at which the system booted, in seconds since
851 The "processes" line gives the number of processes and threads created, which
852 includes (but is not limited to) those created by calls to the fork() and
853 clone() system calls.
855 The "procs_running" line gives the number of processes currently running on
858 The "procs_blocked" line gives the number of processes currently blocked,
859 waiting for I/O to complete.
861 1.9 Ext4 file system parameters
862 ------------------------------
863 Ext4 file system have one directory per partition under /proc/fs/ext4/
864 # ls /proc/fs/ext4/hdc/
865 group_prealloc max_to_scan mb_groups mb_history min_to_scan order2_req
869 This file gives the details of mutiblock allocator buddy cache of free blocks
872 Multiblock allocation history.
875 This file indicate whether the multiblock allocator should start collecting
876 statistics. The statistics are shown during unmount
879 The multiblock allocator normalize the block allocation request to
880 group_prealloc filesystem blocks if we don't have strip value set.
881 The stripe value can be specified at mount time or during mke2fs.
884 How long multiblock allocator can look for a best extent (in found extents)
887 How long multiblock allocator must look for a best extent
890 Multiblock allocator use 2^N search using buddies only for requests greater
891 than or equal to order2_req. The request size is specfied in file system
892 blocks. A value of 2 indicate only if the requests are greater than or equal
896 Files smaller than stream_req are served by the stream allocator, whose
897 purpose is to pack requests as close each to other as possible to
898 produce smooth I/O traffic. Avalue of 16 indicate that file smaller than 16
899 filesystem block size will use group based preallocation.
901 ------------------------------------------------------------------------------
903 ------------------------------------------------------------------------------
904 The /proc file system serves information about the running system. It not only
905 allows access to process data but also allows you to request the kernel status
906 by reading files in the hierarchy.
908 The directory structure of /proc reflects the types of information and makes
909 it easy, if not obvious, where to look for specific data.
910 ------------------------------------------------------------------------------
912 ------------------------------------------------------------------------------
913 CHAPTER 2: MODIFYING SYSTEM PARAMETERS
914 ------------------------------------------------------------------------------
916 ------------------------------------------------------------------------------
918 ------------------------------------------------------------------------------
919 * Modifying kernel parameters by writing into files found in /proc/sys
920 * Exploring the files which modify certain parameters
921 * Review of the /proc/sys file tree
922 ------------------------------------------------------------------------------
925 A very interesting part of /proc is the directory /proc/sys. This is not only
926 a source of information, it also allows you to change parameters within the
927 kernel. Be very careful when attempting this. You can optimize your system,
928 but you can also cause it to crash. Never alter kernel parameters on a
929 production system. Set up a development machine and test to make sure that
930 everything works the way you want it to. You may have no alternative but to
931 reboot the machine once an error has been made.
933 To change a value, simply echo the new value into the file. An example is
934 given below in the section on the file system data. You need to be root to do
935 this. You can create your own boot script to perform this every time your
938 The files in /proc/sys can be used to fine tune and monitor miscellaneous and
939 general things in the operation of the Linux kernel. Since some of the files
940 can inadvertently disrupt your system, it is advisable to read both
941 documentation and source before actually making adjustments. In any case, be
942 very careful when writing to any of these files. The entries in /proc may
943 change slightly between the 2.1.* and the 2.2 kernel, so if there is any doubt
944 review the kernel documentation in the directory /usr/src/linux/Documentation.
945 This chapter is heavily based on the documentation included in the pre 2.2
946 kernels, and became part of it in version 2.2.1 of the Linux kernel.
948 2.1 /proc/sys/fs - File system data
949 -----------------------------------
951 This subdirectory contains specific file system, file handle, inode, dentry
952 and quota information.
954 Currently, these files are in /proc/sys/fs:
959 Status of the directory cache. Since directory entries are dynamically
960 allocated and deallocated, this file indicates the current status. It holds
961 six values, in which the last two are not used and are always zero. The others
962 are listed in table 2-1.
965 Table 2-1: Status files of the directory cache
966 ..............................................................................
968 nr_dentry Almost always zero
969 nr_unused Number of unused cache entries
971 in seconds after the entry may be reclaimed, when memory is short
972 want_pages internally
973 ..............................................................................
975 dquot-nr and dquot-max
976 ----------------------
978 The file dquot-max shows the maximum number of cached disk quota entries.
980 The file dquot-nr shows the number of allocated disk quota entries and the
981 number of free disk quota entries.
983 If the number of available cached disk quotas is very low and you have a large
984 number of simultaneous system users, you might want to raise the limit.
989 The kernel allocates file handles dynamically, but doesn't free them again at
992 The value in file-max denotes the maximum number of file handles that the
993 Linux kernel will allocate. When you get a lot of error messages about running
994 out of file handles, you might want to raise this limit. The default value is
995 10% of RAM in kilobytes. To change it, just write the new number into the
998 # cat /proc/sys/fs/file-max
1000 # echo 8192 > /proc/sys/fs/file-max
1001 # cat /proc/sys/fs/file-max
1005 This method of revision is useful for all customizable parameters of the
1006 kernel - simply echo the new value to the corresponding file.
1008 Historically, the three values in file-nr denoted the number of allocated file
1009 handles, the number of allocated but unused file handles, and the maximum
1010 number of file handles. Linux 2.6 always reports 0 as the number of free file
1011 handles -- this is not an error, it just means that the number of allocated
1012 file handles exactly matches the number of used file handles.
1014 Attempts to allocate more file descriptors than file-max are reported with
1015 printk, look for "VFS: file-max limit <number> reached".
1017 inode-state and inode-nr
1018 ------------------------
1020 The file inode-nr contains the first two items from inode-state, so we'll skip
1023 inode-state contains two actual numbers and five dummy values. The numbers
1024 are nr_inodes and nr_free_inodes (in order of appearance).
1029 Denotes the number of inodes the system has allocated. This number will
1030 grow and shrink dynamically.
1035 Denotes the maximum number of file-handles a process can
1036 allocate. Default value is 1024*1024 (1048576) which should be
1037 enough for most machines. Actual limit depends on RLIMIT_NOFILE
1043 Represents the number of free inodes. Ie. The number of inuse inodes is
1044 (nr_inodes - nr_free_inodes).
1046 aio-nr and aio-max-nr
1047 ---------------------
1049 aio-nr is the running total of the number of events specified on the
1050 io_setup system call for all currently active aio contexts. If aio-nr
1051 reaches aio-max-nr then io_setup will fail with EAGAIN. Note that
1052 raising aio-max-nr does not result in the pre-allocation or re-sizing
1053 of any kernel data structures.
1055 2.2 /proc/sys/fs/binfmt_misc - Miscellaneous binary formats
1056 -----------------------------------------------------------
1058 Besides these files, there is the subdirectory /proc/sys/fs/binfmt_misc. This
1059 handles the kernel support for miscellaneous binary formats.
1061 Binfmt_misc provides the ability to register additional binary formats to the
1062 Kernel without compiling an additional module/kernel. Therefore, binfmt_misc
1063 needs to know magic numbers at the beginning or the filename extension of the
1066 It works by maintaining a linked list of structs that contain a description of
1067 a binary format, including a magic with size (or the filename extension),
1068 offset and mask, and the interpreter name. On request it invokes the given
1069 interpreter with the original program as argument, as binfmt_java and
1070 binfmt_em86 and binfmt_mz do. Since binfmt_misc does not define any default
1071 binary-formats, you have to register an additional binary-format.
1073 There are two general files in binfmt_misc and one file per registered format.
1074 The two general files are register and status.
1076 Registering a new binary format
1077 -------------------------------
1079 To register a new binary format you have to issue the command
1081 echo :name:type:offset:magic:mask:interpreter: > /proc/sys/fs/binfmt_misc/register
1085 with appropriate name (the name for the /proc-dir entry), offset (defaults to
1086 0, if omitted), magic, mask (which can be omitted, defaults to all 0xff) and
1087 last but not least, the interpreter that is to be invoked (for example and
1088 testing /bin/echo). Type can be M for usual magic matching or E for filename
1089 extension matching (give extension in place of magic).
1091 Check or reset the status of the binary format handler
1092 ------------------------------------------------------
1094 If you do a cat on the file /proc/sys/fs/binfmt_misc/status, you will get the
1095 current status (enabled/disabled) of binfmt_misc. Change the status by echoing
1096 0 (disables) or 1 (enables) or -1 (caution: this clears all previously
1097 registered binary formats) to status. For example echo 0 > status to disable
1098 binfmt_misc (temporarily).
1100 Status of a single handler
1101 --------------------------
1103 Each registered handler has an entry in /proc/sys/fs/binfmt_misc. These files
1104 perform the same function as status, but their scope is limited to the actual
1105 binary format. By cating this file, you also receive all related information
1106 about the interpreter/magic of the binfmt.
1108 Example usage of binfmt_misc (emulate binfmt_java)
1109 --------------------------------------------------
1111 cd /proc/sys/fs/binfmt_misc
1112 echo ':Java:M::\xca\xfe\xba\xbe::/usr/local/java/bin/javawrapper:' > register
1113 echo ':HTML:E::html::/usr/local/java/bin/appletviewer:' > register
1114 echo ':Applet:M::<!--applet::/usr/local/java/bin/appletviewer:' > register
1115 echo ':DEXE:M::\x0eDEX::/usr/bin/dosexec:' > register
1118 These four lines add support for Java executables and Java applets (like
1119 binfmt_java, additionally recognizing the .html extension with no need to put
1120 <!--applet> to every applet file). You have to install the JDK and the
1121 shell-script /usr/local/java/bin/javawrapper too. It works around the
1122 brokenness of the Java filename handling. To add a Java binary, just create a
1123 link to the class-file somewhere in the path.
1125 2.3 /proc/sys/kernel - general kernel parameters
1126 ------------------------------------------------
1128 This directory reflects general kernel behaviors. As I've said before, the
1129 contents depend on your configuration. Here you'll find the most important
1130 files, along with descriptions of what they mean and how to use them.
1135 The file contains three values; highwater, lowwater, and frequency.
1137 It exists only when BSD-style process accounting is enabled. These values
1138 control its behavior. If the free space on the file system where the log lives
1139 goes below lowwater percentage, accounting suspends. If it goes above
1140 highwater percentage, accounting resumes. Frequency determines how often you
1141 check the amount of free space (value is in seconds). Default settings are: 4,
1142 2, and 30. That is, suspend accounting if there is less than 2 percent free;
1143 resume it if we have a value of 3 or more percent; consider information about
1144 the amount of free space valid for 30 seconds
1149 When the value in this file is 0, ctrl-alt-del is trapped and sent to the init
1150 program to handle a graceful restart. However, when the value is greater that
1151 zero, Linux's reaction to this key combination will be an immediate reboot,
1152 without syncing its dirty buffers.
1155 When a program (like dosemu) has the keyboard in raw mode, the
1156 ctrl-alt-del is intercepted by the program before it ever reaches the
1157 kernel tty layer, and it is up to the program to decide what to do with
1160 domainname and hostname
1161 -----------------------
1163 These files can be controlled to set the NIS domainname and hostname of your
1164 box. For the classic darkstar.frop.org a simple:
1166 # echo "darkstar" > /proc/sys/kernel/hostname
1167 # echo "frop.org" > /proc/sys/kernel/domainname
1170 would suffice to set your hostname and NIS domainname.
1172 osrelease, ostype and version
1173 -----------------------------
1175 The names make it pretty obvious what these fields contain:
1177 > cat /proc/sys/kernel/osrelease
1180 > cat /proc/sys/kernel/ostype
1183 > cat /proc/sys/kernel/version
1184 #4 Fri Oct 1 12:41:14 PDT 1999
1187 The files osrelease and ostype should be clear enough. Version needs a little
1188 more clarification. The #4 means that this is the 4th kernel built from this
1189 source base and the date after it indicates the time the kernel was built. The
1190 only way to tune these values is to rebuild the kernel.
1195 The value in this file represents the number of seconds the kernel waits
1196 before rebooting on a panic. When you use the software watchdog, the
1197 recommended setting is 60. If set to 0, the auto reboot after a kernel panic
1198 is disabled, which is the default setting.
1203 The four values in printk denote
1205 * default_message_loglevel,
1206 * minimum_console_loglevel and
1207 * default_console_loglevel
1210 These values influence printk() behavior when printing or logging error
1211 messages, which come from inside the kernel. See syslog(2) for more
1212 information on the different log levels.
1217 Messages with a higher priority than this will be printed to the console.
1219 default_message_level
1220 ---------------------
1222 Messages without an explicit priority will be printed with this priority.
1224 minimum_console_loglevel
1225 ------------------------
1227 Minimum (highest) value to which the console_loglevel can be set.
1229 default_console_loglevel
1230 ------------------------
1232 Default value for console_loglevel.
1237 This file shows the size of the generic SCSI (sg) buffer. At this point, you
1238 can't tune it yet, but you can change it at compile time by editing
1239 include/scsi/sg.h and changing the value of SG_BIG_BUFF.
1241 If you use a scanner with SANE (Scanner Access Now Easy) you might want to set
1242 this to a higher value. Refer to the SANE documentation on this issue.
1247 The location where the modprobe binary is located. The kernel uses this
1248 program to load modules on demand.
1253 The value in this file affects behavior of handling NMI. When the value is
1254 non-zero, unknown NMI is trapped and then panic occurs. At that time, kernel
1255 debugging information is displayed on console.
1257 NMI switch that most IA32 servers have fires unknown NMI up, for example.
1258 If a system hangs up, try pressing the NMI switch.
1263 Enables/Disables the NMI watchdog on x86 systems. When the value is non-zero
1264 the NMI watchdog is enabled and will continuously test all online cpus to
1265 determine whether or not they are still functioning properly.
1267 Because the NMI watchdog shares registers with oprofile, by disabling the NMI
1268 watchdog, oprofile may have more registers to utilize.
1273 Enables/Disables the protection of the per-process proc entries "maps" and
1274 "smaps". When enabled, the contents of these files are visible only to
1275 readers that are allowed to ptrace() the given process.
1278 2.4 /proc/sys/vm - The virtual memory subsystem
1279 -----------------------------------------------
1281 The files in this directory can be used to tune the operation of the virtual
1282 memory (VM) subsystem of the Linux kernel.
1287 Controls the tendency of the kernel to reclaim the memory which is used for
1288 caching of directory and inode objects.
1290 At the default value of vfs_cache_pressure=100 the kernel will attempt to
1291 reclaim dentries and inodes at a "fair" rate with respect to pagecache and
1292 swapcache reclaim. Decreasing vfs_cache_pressure causes the kernel to prefer
1293 to retain dentry and inode caches. Increasing vfs_cache_pressure beyond 100
1294 causes the kernel to prefer to reclaim dentries and inodes.
1296 dirty_background_ratio
1297 ----------------------
1299 Contains, as a percentage of total system memory, the number of pages at which
1300 the pdflush background writeback daemon will start writing out dirty data.
1305 Contains, as a percentage of total system memory, the number of pages at which
1306 a process which is generating disk writes will itself start writing out dirty
1309 dirty_writeback_centisecs
1310 -------------------------
1312 The pdflush writeback daemons will periodically wake up and write `old' data
1313 out to disk. This tunable expresses the interval between those wakeups, in
1314 100'ths of a second.
1316 Setting this to zero disables periodic writeback altogether.
1318 dirty_expire_centisecs
1319 ----------------------
1321 This tunable is used to define when dirty data is old enough to be eligible
1322 for writeout by the pdflush daemons. It is expressed in 100'ths of a second.
1323 Data which has been dirty in-memory for longer than this interval will be
1324 written out next time a pdflush daemon wakes up.
1326 highmem_is_dirtyable
1327 --------------------
1329 Only present if CONFIG_HIGHMEM is set.
1331 This defaults to 0 (false), meaning that the ratios set above are calculated
1332 as a percentage of lowmem only. This protects against excessive scanning
1333 in page reclaim, swapping and general VM distress.
1335 Setting this to 1 can be useful on 32 bit machines where you want to make
1336 random changes within an MMAPed file that is larger than your available
1337 lowmem without causing large quantities of random IO. Is is safe if the
1338 behavior of all programs running on the machine is known and memory will
1339 not be otherwise stressed.
1344 If non-zero, this sysctl disables the new 32-bit mmap mmap layout - the kernel
1345 will use the legacy (2.4) layout for all processes.
1347 lowmem_reserve_ratio
1348 ---------------------
1350 For some specialised workloads on highmem machines it is dangerous for
1351 the kernel to allow process memory to be allocated from the "lowmem"
1352 zone. This is because that memory could then be pinned via the mlock()
1353 system call, or by unavailability of swapspace.
1355 And on large highmem machines this lack of reclaimable lowmem memory
1358 So the Linux page allocator has a mechanism which prevents allocations
1359 which _could_ use highmem from using too much lowmem. This means that
1360 a certain amount of lowmem is defended from the possibility of being
1361 captured into pinned user memory.
1363 (The same argument applies to the old 16 megabyte ISA DMA region. This
1364 mechanism will also defend that region from allocations which could use
1367 The `lowmem_reserve_ratio' tunable determines how aggressive the kernel is
1368 in defending these lower zones.
1370 If you have a machine which uses highmem or ISA DMA and your
1371 applications are using mlock(), or if you are running with no swap then
1372 you probably should change the lowmem_reserve_ratio setting.
1374 The lowmem_reserve_ratio is an array. You can see them by reading this file.
1376 % cat /proc/sys/vm/lowmem_reserve_ratio
1379 Note: # of this elements is one fewer than number of zones. Because the highest
1380 zone's value is not necessary for following calculation.
1382 But, these values are not used directly. The kernel calculates # of protection
1383 pages for each zones from them. These are shown as array of protection pages
1384 in /proc/zoneinfo like followings. (This is an example of x86-64 box).
1385 Each zone has an array of protection pages like this.
1396 protection: (0, 2004, 2004, 2004)
1397 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
1402 These protections are added to score to judge whether this zone should be used
1403 for page allocation or should be reclaimed.
1405 In this example, if normal pages (index=2) are required to this DMA zone and
1406 pages_high is used for watermark, the kernel judges this zone should not be
1407 used because pages_free(1355) is smaller than watermark + protection[2]
1408 (4 + 2004 = 2008). If this protection value is 0, this zone would be used for
1409 normal page requirement. If requirement is DMA zone(index=0), protection[0]
1412 zone[i]'s protection[j] is calculated by following exprssion.
1415 zone[i]->protection[j]
1416 = (total sums of present_pages from zone[i+1] to zone[j] on the node)
1417 / lowmem_reserve_ratio[i];
1419 (should not be protected. = 0;
1421 (not necessary, but looks 0)
1423 The default values of lowmem_reserve_ratio[i] are
1424 256 (if zone[i] means DMA or DMA32 zone)
1426 As above expression, they are reciprocal number of ratio.
1427 256 means 1/256. # of protection pages becomes about "0.39%" of total present
1428 pages of higher zones on the node.
1430 If you would like to protect more pages, smaller values are effective.
1431 The minimum value is 1 (1/1 -> 100%).
1436 page-cluster controls the number of pages which are written to swap in
1437 a single attempt. The swap I/O size.
1439 It is a logarithmic value - setting it to zero means "1 page", setting
1440 it to 1 means "2 pages", setting it to 2 means "4 pages", etc.
1442 The default value is three (eight pages at a time). There may be some
1443 small benefits in tuning this to a different value if your workload is
1449 Controls overcommit of system memory, possibly allowing processes
1450 to allocate (but not use) more memory than is actually available.
1453 0 - Heuristic overcommit handling. Obvious overcommits of
1454 address space are refused. Used for a typical system. It
1455 ensures a seriously wild allocation fails while allowing
1456 overcommit to reduce swap usage. root is allowed to
1457 allocate slightly more memory in this mode. This is the
1460 1 - Always overcommit. Appropriate for some scientific
1463 2 - Don't overcommit. The total address space commit
1464 for the system is not permitted to exceed swap plus a
1465 configurable percentage (default is 50) of physical RAM.
1466 Depending on the percentage you use, in most situations
1467 this means a process will not be killed while attempting
1468 to use already-allocated memory but will receive errors
1469 on memory allocation as appropriate.
1474 Percentage of physical memory size to include in overcommit calculations
1477 Memory allocation limit = swapspace + physmem * (overcommit_ratio / 100)
1479 swapspace = total size of all swap areas
1480 physmem = size of physical memory in system
1482 nr_hugepages and hugetlb_shm_group
1483 ----------------------------------
1485 nr_hugepages configures number of hugetlb page reserved for the system.
1487 hugetlb_shm_group contains group id that is allowed to create SysV shared
1488 memory segment using hugetlb page.
1490 hugepages_treat_as_movable
1491 --------------------------
1493 This parameter is only useful when kernelcore= is specified at boot time to
1494 create ZONE_MOVABLE for pages that may be reclaimed or migrated. Huge pages
1495 are not movable so are not normally allocated from ZONE_MOVABLE. A non-zero
1496 value written to hugepages_treat_as_movable allows huge pages to be allocated
1499 Once enabled, the ZONE_MOVABLE is treated as an area of memory the huge
1500 pages pool can easily grow or shrink within. Assuming that applications are
1501 not running that mlock() a lot of memory, it is likely the huge pages pool
1502 can grow to the size of ZONE_MOVABLE by repeatedly entering the desired value
1503 into nr_hugepages and triggering page reclaim.
1508 laptop_mode is a knob that controls "laptop mode". All the things that are
1509 controlled by this knob are discussed in Documentation/laptops/laptop-mode.txt.
1514 block_dump enables block I/O debugging when set to a nonzero value. More
1515 information on block I/O debugging is in Documentation/laptops/laptop-mode.txt.
1520 This file contains valid hold time of swap out protection token. The Linux
1521 VM has token based thrashing control mechanism and uses the token to prevent
1522 unnecessary page faults in thrashing situation. The unit of the value is
1523 second. The value would be useful to tune thrashing behavior.
1528 Writing to this will cause the kernel to drop clean caches, dentries and
1529 inodes from memory, causing that memory to become free.
1532 echo 1 > /proc/sys/vm/drop_caches
1533 To free dentries and inodes:
1534 echo 2 > /proc/sys/vm/drop_caches
1535 To free pagecache, dentries and inodes:
1536 echo 3 > /proc/sys/vm/drop_caches
1538 As this is a non-destructive operation and dirty objects are not freeable, the
1539 user should run `sync' first.
1542 2.5 /proc/sys/dev - Device specific parameters
1543 ----------------------------------------------
1545 Currently there is only support for CDROM drives, and for those, there is only
1546 one read-only file containing information about the CD-ROM drives attached to
1549 >cat /proc/sys/dev/cdrom/info
1550 CD-ROM information, Id: cdrom.c 2.55 1999/04/25
1554 drive # of slots: 1 0
1558 Can change speed: 1 1
1559 Can select disk: 0 1
1560 Can read multisession: 1 1
1562 Reports media changed: 1 1
1566 You see two drives, sr0 and hdb, along with a list of their features.
1568 2.6 /proc/sys/sunrpc - Remote procedure calls
1569 ---------------------------------------------
1571 This directory contains four files, which enable or disable debugging for the
1572 RPC functions NFS, NFS-daemon, RPC and NLM. The default values are 0. They can
1573 be set to one to turn debugging on. (The default value is 0 for each)
1575 2.7 /proc/sys/net - Networking stuff
1576 ------------------------------------
1578 The interface to the networking parts of the kernel is located in
1579 /proc/sys/net. Table 2-3 shows all possible subdirectories. You may see only
1580 some of them, depending on your kernel's configuration.
1583 Table 2-3: Subdirectories in /proc/sys/net
1584 ..............................................................................
1585 Directory Content Directory Content
1586 core General parameter appletalk Appletalk protocol
1587 unix Unix domain sockets netrom NET/ROM
1588 802 E802 protocol ax25 AX25
1589 ethernet Ethernet protocol rose X.25 PLP layer
1590 ipv4 IP version 4 x25 X.25 protocol
1591 ipx IPX token-ring IBM token ring
1592 bridge Bridging decnet DEC net
1594 ..............................................................................
1596 We will concentrate on IP networking here. Since AX15, X.25, and DEC Net are
1597 only minor players in the Linux world, we'll skip them in this chapter. You'll
1598 find some short info on Appletalk and IPX further on in this chapter. Review
1599 the online documentation and the kernel source to get a detailed view of the
1600 parameters for those protocols. In this section we'll discuss the
1601 subdirectories printed in bold letters in the table above. As default values
1602 are suitable for most needs, there is no need to change these values.
1604 /proc/sys/net/core - Network core options
1605 -----------------------------------------
1610 The default setting of the socket receive buffer in bytes.
1615 The maximum receive socket buffer size in bytes.
1620 The default setting (in bytes) of the socket send buffer.
1625 The maximum send socket buffer size in bytes.
1627 message_burst and message_cost
1628 ------------------------------
1630 These parameters are used to limit the warning messages written to the kernel
1631 log from the networking code. They enforce a rate limit to make a
1632 denial-of-service attack impossible. A higher message_cost factor, results in
1633 fewer messages that will be written. Message_burst controls when messages will
1634 be dropped. The default settings limit warning messages to one every five
1640 This controls console messages from the networking stack that can occur because
1641 of problems on the network like duplicate address or bad checksums. Normally,
1642 this should be enabled, but if the problem persists the messages can be
1649 Maximum number of packets, queued on the INPUT side, when the interface
1650 receives packets faster than kernel can process them.
1655 Maximum ancillary buffer size allowed per socket. Ancillary data is a sequence
1656 of struct cmsghdr structures with appended data.
1658 /proc/sys/net/unix - Parameters for Unix domain sockets
1659 -------------------------------------------------------
1661 There are only two files in this subdirectory. They control the delays for
1662 deleting and destroying socket descriptors.
1664 2.8 /proc/sys/net/ipv4 - IPV4 settings
1665 --------------------------------------
1667 IP version 4 is still the most used protocol in Unix networking. It will be
1668 replaced by IP version 6 in the next couple of years, but for the moment it's
1669 the de facto standard for the internet and is used in most networking
1670 environments around the world. Because of the importance of this protocol,
1671 we'll have a deeper look into the subtree controlling the behavior of the IPv4
1672 subsystem of the Linux kernel.
1674 Let's start with the entries in /proc/sys/net/ipv4.
1679 icmp_echo_ignore_all and icmp_echo_ignore_broadcasts
1680 ----------------------------------------------------
1682 Turn on (1) or off (0), if the kernel should ignore all ICMP ECHO requests, or
1683 just those to broadcast and multicast addresses.
1685 Please note that if you accept ICMP echo requests with a broadcast/multi\-cast
1686 destination address your network may be used as an exploder for denial of
1687 service packet flooding attacks to other hosts.
1689 icmp_destunreach_rate, icmp_echoreply_rate, icmp_paramprob_rate and icmp_timeexeed_rate
1690 ---------------------------------------------------------------------------------------
1692 Sets limits for sending ICMP packets to specific targets. A value of zero
1693 disables all limiting. Any positive value sets the maximum package rate in
1694 hundredth of a second (on Intel systems).
1702 This file contains the number one if the host received its IP configuration by
1703 RARP, BOOTP, DHCP or a similar mechanism. Otherwise it is zero.
1708 TTL (Time To Live) for IPv4 interfaces. This is simply the maximum number of
1709 hops a packet may travel.
1714 Enable dynamic socket address rewriting on interface address change. This is
1715 useful for dialup interface with changing IP addresses.
1720 Enable or disable forwarding of IP packages between interfaces. Changing this
1721 value resets all other parameters to their default values. They differ if the
1722 kernel is configured as host or router.
1727 Range of ports used by TCP and UDP to choose the local port. Contains two
1728 numbers, the first number is the lowest port, the second number the highest
1729 local port. Default is 1024-4999. Should be changed to 32768-61000 for
1735 Global switch to turn path MTU discovery off. It can also be set on a per
1736 socket basis by the applications or on a per route basis.
1741 Enable/disable debugging of IP masquerading.
1743 IP fragmentation settings
1744 -------------------------
1746 ipfrag_high_trash and ipfrag_low_trash
1747 --------------------------------------
1749 Maximum memory used to reassemble IP fragments. When ipfrag_high_thresh bytes
1750 of memory is allocated for this purpose, the fragment handler will toss
1751 packets until ipfrag_low_thresh is reached.
1756 Time in seconds to keep an IP fragment in memory.
1764 This file controls the use of the ECN bit in the IPv4 headers. This is a new
1765 feature about Explicit Congestion Notification, but some routers and firewalls
1766 block traffic that has this bit set, so it could be necessary to echo 0 to
1767 /proc/sys/net/ipv4/tcp_ecn if you want to talk to these sites. For more info
1768 you could read RFC2481.
1770 tcp_retrans_collapse
1771 --------------------
1773 Bug-to-bug compatibility with some broken printers. On retransmit, try to send
1774 larger packets to work around bugs in certain TCP stacks. Can be turned off by
1777 tcp_keepalive_probes
1778 --------------------
1780 Number of keep alive probes TCP sends out, until it decides that the
1781 connection is broken.
1786 How often TCP sends out keep alive messages, when keep alive is enabled. The
1792 Number of times initial SYNs for a TCP connection attempt will be
1793 retransmitted. Should not be higher than 255. This is only the timeout for
1794 outgoing connections, for incoming connections the number of retransmits is
1795 defined by tcp_retries1.
1800 Enable select acknowledgments after RFC2018.
1805 Enable timestamps as defined in RFC1323.
1810 Enable the strict RFC793 interpretation of the TCP urgent pointer field. The
1811 default is to use the BSD compatible interpretation of the urgent pointer
1812 pointing to the first byte after the urgent data. The RFC793 interpretation is
1813 to have it point to the last byte of urgent data. Enabling this option may
1814 lead to interoperability problems. Disabled by default.
1819 Only valid when the kernel was compiled with CONFIG_SYNCOOKIES. Send out
1820 syncookies when the syn backlog queue of a socket overflows. This is to ward
1821 off the common 'syn flood attack'. Disabled by default.
1823 Note that the concept of a socket backlog is abandoned. This means the peer
1824 may not receive reliable error messages from an over loaded server with
1830 Enable window scaling as defined in RFC1323.
1835 The length of time in seconds it takes to receive a final FIN before the
1836 socket is always closed. This is strictly a violation of the TCP
1837 specification, but required to prevent denial-of-service attacks.
1842 Indicates how many keep alive probes are sent per slow timer run. Should not
1843 be set too high to prevent bursts.
1848 Length of the per socket backlog queue. Since Linux 2.2 the backlog specified
1849 in listen(2) only specifies the length of the backlog queue of already
1850 established sockets. When more connection requests arrive Linux starts to drop
1851 packets. When syncookies are enabled the packets are still answered and the
1852 maximum queue is effectively ignored.
1857 Defines how often an answer to a TCP connection request is retransmitted
1863 Defines how often a TCP packet is retransmitted before giving up.
1865 Interface specific settings
1866 ---------------------------
1868 In the directory /proc/sys/net/ipv4/conf you'll find one subdirectory for each
1869 interface the system knows about and one directory calls all. Changes in the
1870 all subdirectory affect all interfaces, whereas changes in the other
1871 subdirectories affect only one interface. All directories have the same
1877 This switch decides if the kernel accepts ICMP redirect messages or not. The
1878 default is 'yes' if the kernel is configured for a regular host and 'no' for a
1879 router configuration.
1884 Should source routed packages be accepted or declined. The default is
1885 dependent on the kernel configuration. It's 'yes' for routers and 'no' for
1891 Accept packets with source address 0.b.c.d with destinations not to this host
1892 as local ones. It is supposed that a BOOTP relay daemon will catch and forward
1895 The default is 0, since this feature is not implemented yet (kernel version
1901 Enable or disable IP forwarding on this interface.
1906 Log packets with source addresses with no known route to kernel log.
1911 Do multicast routing. The kernel needs to be compiled with CONFIG_MROUTE and a
1912 multicast routing daemon is required.
1917 Does (1) or does not (0) perform proxy ARP.
1922 Integer value determines if a source validation should be made. 1 means yes, 0
1923 means no. Disabled by default, but local/broadcast address spoofing is always
1926 If you set this to 1 on a router that is the only connection for a network to
1927 the net, it will prevent spoofing attacks against your internal networks
1928 (external addresses can still be spoofed), without the need for additional
1934 Accept ICMP redirect messages only for gateways, listed in default gateway
1935 list. Enabled by default.
1940 If it is not set the kernel does not assume that different subnets on this
1941 device can communicate directly. Default setting is 'yes'.
1946 Determines whether to send ICMP redirects to other hosts.
1951 The directory /proc/sys/net/ipv4/route contains several file to control
1954 error_burst and error_cost
1955 --------------------------
1957 These parameters are used to limit how many ICMP destination unreachable to
1958 send from the host in question. ICMP destination unreachable messages are
1959 sent when we cannot reach the next hop while trying to transmit a packet.
1960 It will also print some error messages to kernel logs if someone is ignoring
1961 our ICMP redirects. The higher the error_cost factor is, the fewer
1962 destination unreachable and error messages will be let through. Error_burst
1963 controls when destination unreachable messages and error messages will be
1964 dropped. The default settings limit warning messages to five every second.
1969 Writing to this file results in a flush of the routing cache.
1971 gc_elasticity, gc_interval, gc_min_interval_ms, gc_timeout, gc_thresh
1972 ---------------------------------------------------------------------
1974 Values to control the frequency and behavior of the garbage collection
1975 algorithm for the routing cache. gc_min_interval is deprecated and replaced
1976 by gc_min_interval_ms.
1982 Maximum size of the routing cache. Old entries will be purged once the cache
1983 reached has this size.
1985 redirect_load, redirect_number
1986 ------------------------------
1988 Factors which determine if more ICPM redirects should be sent to a specific
1989 host. No redirects will be sent once the load limit or the maximum number of
1990 redirects has been reached.
1995 Timeout for redirects. After this period redirects will be sent again, even if
1996 this has been stopped, because the load or number limit has been reached.
1998 Network Neighbor handling
1999 -------------------------
2001 Settings about how to handle connections with direct neighbors (nodes attached
2002 to the same link) can be found in the directory /proc/sys/net/ipv4/neigh.
2004 As we saw it in the conf directory, there is a default subdirectory which
2005 holds the default values, and one directory for each interface. The contents
2006 of the directories are identical, with the single exception that the default
2007 settings contain additional options to set garbage collection parameters.
2009 In the interface directories you'll find the following entries:
2011 base_reachable_time, base_reachable_time_ms
2012 -------------------------------------------
2014 A base value used for computing the random reachable time value as specified
2017 Expression of base_reachable_time, which is deprecated, is in seconds.
2018 Expression of base_reachable_time_ms is in milliseconds.
2020 retrans_time, retrans_time_ms
2021 -----------------------------
2023 The time between retransmitted Neighbor Solicitation messages.
2024 Used for address resolution and to determine if a neighbor is
2027 Expression of retrans_time, which is deprecated, is in 1/100 seconds (for
2028 IPv4) or in jiffies (for IPv6).
2029 Expression of retrans_time_ms is in milliseconds.
2034 Maximum queue length for a pending arp request - the number of packets which
2035 are accepted from other layers while the ARP address is still resolved.
2040 Maximum for random delay of answers to neighbor solicitation messages in
2041 jiffies (1/100 sec). Not yet implemented (Linux does not have anycast support
2047 Maximum number of retries for unicast solicitation.
2052 Maximum number of retries for multicast solicitation.
2054 delay_first_probe_time
2055 ----------------------
2057 Delay for the first time probe if the neighbor is reachable. (see
2063 An ARP/neighbor entry is only replaced with a new one if the old is at least
2064 locktime old. This prevents ARP cache thrashing.
2069 Maximum time (real time is random [0..proxytime]) before answering to an ARP
2070 request for which we have an proxy ARP entry. In some cases, this is used to
2071 prevent network flooding.
2076 Maximum queue length of the delayed proxy arp timer. (see proxy_delay).
2081 Determines the number of requests to send to the user level ARP daemon. Use 0
2087 Determines how often to check for stale ARP entries. After an ARP entry is
2088 stale it will be resolved again (which is useful when an IP address migrates
2089 to another machine). When ucast_solicit is greater than 0 it first tries to
2090 send an ARP packet directly to the known host When that fails and
2091 mcast_solicit is greater than 0, an ARP request is broadcasted.
2096 The /proc/sys/net/appletalk directory holds the Appletalk configuration data
2097 when Appletalk is loaded. The configurable parameters are:
2102 The amount of time we keep an ARP entry before expiring it. Used to age out
2108 The amount of time we will spend trying to resolve an Appletalk address.
2110 aarp-retransmit-limit
2111 ---------------------
2113 The number of times we will retransmit a query before giving up.
2118 Controls the rate at which expires are checked.
2120 The directory /proc/net/appletalk holds the list of active Appletalk sockets
2123 The fields indicate the DDP type, the local address (in network:node format)
2124 the remote address, the size of the transmit pending queue, the size of the
2125 received queue (bytes waiting for applications to read) the state and the uid
2128 /proc/net/atalk_iface lists all the interfaces configured for appletalk.It
2129 shows the name of the interface, its Appletalk address, the network range on
2130 that address (or network number for phase 1 networks), and the status of the
2133 /proc/net/atalk_route lists each known network route. It lists the target
2134 (network) that the route leads to, the router (may be directly connected), the
2135 route flags, and the device the route is using.
2140 The IPX protocol has no tunable values in proc/sys/net.
2142 The IPX protocol does, however, provide proc/net/ipx. This lists each IPX
2143 socket giving the local and remote addresses in Novell format (that is
2144 network:node:port). In accordance with the strange Novell tradition,
2145 everything but the port is in hex. Not_Connected is displayed for sockets that
2146 are not tied to a specific remote address. The Tx and Rx queue sizes indicate
2147 the number of bytes pending for transmission and reception. The state
2148 indicates the state the socket is in and the uid is the owning uid of the
2151 The /proc/net/ipx_interface file lists all IPX interfaces. For each interface
2152 it gives the network number, the node number, and indicates if the network is
2153 the primary network. It also indicates which device it is bound to (or
2154 Internal for internal networks) and the Frame Type if appropriate. Linux
2155 supports 802.3, 802.2, 802.2 SNAP and DIX (Blue Book) ethernet framing for
2158 The /proc/net/ipx_route table holds a list of IPX routes. For each route it
2159 gives the destination network, the router node (or Directly) and the network
2160 address of the router (or Connected) for internal networks.
2162 2.11 /proc/sys/fs/mqueue - POSIX message queues filesystem
2163 ----------------------------------------------------------
2165 The "mqueue" filesystem provides the necessary kernel features to enable the
2166 creation of a user space library that implements the POSIX message queues
2167 API (as noted by the MSG tag in the POSIX 1003.1-2001 version of the System
2168 Interfaces specification.)
2170 The "mqueue" filesystem contains values for determining/setting the amount of
2171 resources used by the file system.
2173 /proc/sys/fs/mqueue/queues_max is a read/write file for setting/getting the
2174 maximum number of message queues allowed on the system.
2176 /proc/sys/fs/mqueue/msg_max is a read/write file for setting/getting the
2177 maximum number of messages in a queue value. In fact it is the limiting value
2178 for another (user) limit which is set in mq_open invocation. This attribute of
2179 a queue must be less or equal then msg_max.
2181 /proc/sys/fs/mqueue/msgsize_max is a read/write file for setting/getting the
2182 maximum message size value (it is every message queue's attribute set during
2185 2.12 /proc/<pid>/oom_adj - Adjust the oom-killer score
2186 ------------------------------------------------------
2188 This file can be used to adjust the score used to select which processes
2189 should be killed in an out-of-memory situation. Giving it a high score will
2190 increase the likelihood of this process being killed by the oom-killer. Valid
2191 values are in the range -16 to +15, plus the special value -17, which disables
2192 oom-killing altogether for this process.
2194 2.13 /proc/<pid>/oom_score - Display current oom-killer score
2195 -------------------------------------------------------------
2197 ------------------------------------------------------------------------------
2198 This file can be used to check the current score used by the oom-killer is for
2199 any given <pid>. Use it together with /proc/<pid>/oom_adj to tune which
2200 process should be killed in an out-of-memory situation.
2202 ------------------------------------------------------------------------------
2204 ------------------------------------------------------------------------------
2205 Certain aspects of kernel behavior can be modified at runtime, without the
2206 need to recompile the kernel, or even to reboot the system. The files in the
2207 /proc/sys tree can not only be read, but also modified. You can use the echo
2208 command to write value into these files, thereby changing the default settings
2210 ------------------------------------------------------------------------------
2212 2.14 /proc/<pid>/io - Display the IO accounting fields
2213 -------------------------------------------------------
2215 This file contains IO statistics for each running process
2220 test:/tmp # dd if=/dev/zero of=/tmp/test.dat &
2223 test:/tmp # cat /proc/3828/io
2229 write_bytes: 323932160
2230 cancelled_write_bytes: 0
2239 I/O counter: chars read
2240 The number of bytes which this task has caused to be read from storage. This
2241 is simply the sum of bytes which this process passed to read() and pread().
2242 It includes things like tty IO and it is unaffected by whether or not actual
2243 physical disk IO was required (the read might have been satisfied from
2250 I/O counter: chars written
2251 The number of bytes which this task has caused, or shall cause to be written
2252 to disk. Similar caveats apply here as with rchar.
2258 I/O counter: read syscalls
2259 Attempt to count the number of read I/O operations, i.e. syscalls like read()
2266 I/O counter: write syscalls
2267 Attempt to count the number of write I/O operations, i.e. syscalls like
2268 write() and pwrite().
2274 I/O counter: bytes read
2275 Attempt to count the number of bytes which this process really did cause to
2276 be fetched from the storage layer. Done at the submit_bio() level, so it is
2277 accurate for block-backed filesystems. <please add status regarding NFS and
2278 CIFS at a later time>
2284 I/O counter: bytes written
2285 Attempt to count the number of bytes which this process caused to be sent to
2286 the storage layer. This is done at page-dirtying time.
2289 cancelled_write_bytes
2290 ---------------------
2292 The big inaccuracy here is truncate. If a process writes 1MB to a file and
2293 then deletes the file, it will in fact perform no writeout. But it will have
2294 been accounted as having caused 1MB of write.
2295 In other words: The number of bytes which this process caused to not happen,
2296 by truncating pagecache. A task can cause "negative" IO too. If this task
2297 truncates some dirty pagecache, some IO which another task has been accounted
2298 for (in it's write_bytes) will not be happening. We _could_ just subtract that
2299 from the truncating task's write_bytes, but there is information loss in doing
2306 At its current implementation state, this is a bit racy on 32-bit machines: if
2307 process A reads process B's /proc/pid/io while process B is updating one of
2308 those 64-bit counters, process A could see an intermediate result.
2311 More information about this can be found within the taskstats documentation in
2312 Documentation/accounting.
2314 2.15 /proc/<pid>/coredump_filter - Core dump filtering settings
2315 ---------------------------------------------------------------
2316 When a process is dumped, all anonymous memory is written to a core file as
2317 long as the size of the core file isn't limited. But sometimes we don't want
2318 to dump some memory segments, for example, huge shared memory. Conversely,
2319 sometimes we want to save file-backed memory segments into a core file, not
2320 only the individual files.
2322 /proc/<pid>/coredump_filter allows you to customize which memory segments
2323 will be dumped when the <pid> process is dumped. coredump_filter is a bitmask
2324 of memory types. If a bit of the bitmask is set, memory segments of the
2325 corresponding memory type are dumped, otherwise they are not dumped.
2327 The following 4 memory types are supported:
2328 - (bit 0) anonymous private memory
2329 - (bit 1) anonymous shared memory
2330 - (bit 2) file-backed private memory
2331 - (bit 3) file-backed shared memory
2333 Note that MMIO pages such as frame buffer are never dumped and vDSO pages
2334 are always dumped regardless of the bitmask status.
2336 Default value of coredump_filter is 0x3; this means all anonymous memory
2337 segments are dumped.
2339 If you don't want to dump all shared memory segments attached to pid 1234,
2340 write 1 to the process's proc file.
2342 $ echo 0x1 > /proc/1234/coredump_filter
2344 When a new process is created, the process inherits the bitmask status from its
2345 parent. It is useful to set up coredump_filter before the program runs.
2348 $ echo 0x7 > /proc/self/coredump_filter
2351 ------------------------------------------------------------------------------
projects / linux-2.6 / blob