| blob | 9fb6cbe70bdefdd1a753795cc57ed40019ef7267 |
1 ------------------------------------------------------------------------------
2 T H E /proc F I L E S Y S T E M
3 ------------------------------------------------------------------------------
4 /proc/sys Terrehon Bowden <terrehon@pacbell.net> October 7 1999
5 Bodo Bauer <bb@ricochet.net>
7 2.4.x update Jorge Nerin <comandante@zaralinux.com> November 14 2000
8 move /proc/sys Shen Feng <shen@cn.fujitsu.com> April 1 2009
9 ------------------------------------------------------------------------------
10 Version 1.3 Kernel version 2.2.12
11 Kernel version 2.4.0-test11-pre4
12 ------------------------------------------------------------------------------
13 fixes/update part 1.1 Stefani Seibold <stefani@seibold.net> June 9 2009
15 Table of Contents
16 -----------------
18 0 Preface
19 0.1 Introduction/Credits
20 0.2 Legal Stuff
22 1 Collecting System Information
23 1.1 Process-Specific Subdirectories
24 1.2 Kernel data
25 1.3 IDE devices in /proc/ide
26 1.4 Networking info in /proc/net
27 1.5 SCSI info
28 1.6 Parallel port info in /proc/parport
29 1.7 TTY info in /proc/tty
30 1.8 Miscellaneous kernel statistics in /proc/stat
31 1.9 Ext4 file system parameters
33 2 Modifying System Parameters
35 3 Per-Process Parameters
36 3.1 /proc/<pid>/oom_adj - Adjust the oom-killer score
37 3.2 /proc/<pid>/oom_score - Display current oom-killer score
38 3.3 /proc/<pid>/io - Display the IO accounting fields
39 3.4 /proc/<pid>/coredump_filter - Core dump filtering settings
40 3.5 /proc/<pid>/mountinfo - Information about mounts
41 3.6 /proc/<pid>/comm & /proc/<pid>/task/<tid>/comm
44 ------------------------------------------------------------------------------
45 Preface
46 ------------------------------------------------------------------------------
48 0.1 Introduction/Credits
49 ------------------------
51 This documentation is part of a soon (or so we hope) to be released book on
52 the SuSE Linux distribution. As there is no complete documentation for the
53 /proc file system and we've used many freely available sources to write these
54 chapters, it seems only fair to give the work back to the Linux community.
55 This work is based on the 2.2.* kernel version and the upcoming 2.4.*. I'm
56 afraid it's still far from complete, but we hope it will be useful. As far as
57 we know, it is the first 'all-in-one' document about the /proc file system. It
58 is focused on the Intel x86 hardware, so if you are looking for PPC, ARM,
59 SPARC, AXP, etc., features, you probably won't find what you are looking for.
60 It also only covers IPv4 networking, not IPv6 nor other protocols - sorry. But
61 additions and patches are welcome and will be added to this document if you
62 mail them to Bodo.
64 We'd like to thank Alan Cox, Rik van Riel, and Alexey Kuznetsov and a lot of
65 other people for help compiling this documentation. We'd also like to extend a
66 special thank you to Andi Kleen for documentation, which we relied on heavily
67 to create this document, as well as the additional information he provided.
68 Thanks to everybody else who contributed source or docs to the Linux kernel
69 and helped create a great piece of software... :)
71 If you have any comments, corrections or additions, please don't hesitate to
72 contact Bodo Bauer at bb@ricochet.net. We'll be happy to add them to this
73 document.
75 The latest version of this document is available online at
76 http://skaro.nightcrawler.com/~bb/Docs/Proc as HTML version.
78 If the above direction does not works for you, ypu could try the kernel
79 mailing list at linux-kernel@vger.kernel.org and/or try to reach me at
80 comandante@zaralinux.com.
82 0.2 Legal Stuff
83 ---------------
85 We don't guarantee the correctness of this document, and if you come to us
86 complaining about how you screwed up your system because of incorrect
87 documentation, we won't feel responsible...
89 ------------------------------------------------------------------------------
90 CHAPTER 1: COLLECTING SYSTEM INFORMATION
91 ------------------------------------------------------------------------------
93 ------------------------------------------------------------------------------
94 In This Chapter
95 ------------------------------------------------------------------------------
96 * Investigating the properties of the pseudo file system /proc and its
97 ability to provide information on the running Linux system
98 * Examining /proc's structure
99 * Uncovering various information about the kernel and the processes running
100 on the system
101 ------------------------------------------------------------------------------
104 The proc file system acts as an interface to internal data structures in the
105 kernel. It can be used to obtain information about the system and to change
106 certain kernel parameters at runtime (sysctl).
108 First, we'll take a look at the read-only parts of /proc. In Chapter 2, we
109 show you how you can use /proc/sys to change settings.
111 1.1 Process-Specific Subdirectories
112 -----------------------------------
114 The directory /proc contains (among other things) one subdirectory for each
115 process running on the system, which is named after the process ID (PID).
117 The link self points to the process reading the file system. Each process
118 subdirectory has the entries listed in Table 1-1.
121 Table 1-1: Process specific entries in /proc
122 ..............................................................................
123 File Content
124 clear_refs Clears page referenced bits shown in smaps output
125 cmdline Command line arguments
126 cpu Current and last cpu in which it was executed (2.4)(smp)
127 cwd Link to the current working directory
128 environ Values of environment variables
129 exe Link to the executable of this process
130 fd Directory, which contains all file descriptors
131 maps Memory maps to executables and library files (2.4)
132 mem Memory held by this process
133 root Link to the root directory of this process
134 stat Process status
135 statm Process memory status information
136 status Process status in human readable form
137 wchan If CONFIG_KALLSYMS is set, a pre-decoded wchan
138 stack Report full stack trace, enable via CONFIG_STACKTRACE
139 smaps a extension based on maps, showing the memory consumption of
140 each mapping
141 ..............................................................................
143 For example, to get the status information of a process, all you have to do is
144 read the file /proc/PID/status:
146 >cat /proc/self/status
147 Name: cat
148 State: R (running)
149 Tgid: 5452
150 Pid: 5452
151 PPid: 743
152 TracerPid: 0 (2.4)
153 Uid: 501 501 501 501
154 Gid: 100 100 100 100
155 FDSize: 256
156 Groups: 100 14 16
157 VmPeak: 5004 kB
158 VmSize: 5004 kB
159 VmLck: 0 kB
160 VmHWM: 476 kB
161 VmRSS: 476 kB
162 VmData: 156 kB
163 VmStk: 88 kB
164 VmExe: 68 kB
165 VmLib: 1412 kB
166 VmPTE: 20 kb
167 VmSwap: 0 kB
168 Threads: 1
169 SigQ: 0/28578
170 SigPnd: 0000000000000000
171 ShdPnd: 0000000000000000
172 SigBlk: 0000000000000000
173 SigIgn: 0000000000000000
174 SigCgt: 0000000000000000
175 CapInh: 00000000fffffeff
176 CapPrm: 0000000000000000
177 CapEff: 0000000000000000
178 CapBnd: ffffffffffffffff
179 voluntary_ctxt_switches: 0
180 nonvoluntary_ctxt_switches: 1
182 This shows you nearly the same information you would get if you viewed it with
183 the ps command. In fact, ps uses the proc file system to obtain its
184 information. But you get a more detailed view of the process by reading the
185 file /proc/PID/status. It fields are described in table 1-2.
187 The statm file contains more detailed information about the process
188 memory usage. Its seven fields are explained in Table 1-3. The stat file
189 contains details information about the process itself. Its fields are
190 explained in Table 1-4.
192 (for SMP CONFIG users)
193 For making accounting scalable, RSS related information are handled in
194 asynchronous manner and the vaule may not be very precise. To see a precise
195 snapshot of a moment, you can see /proc/<pid>/smaps file and scan page table.
196 It's slow but very precise.
198 Table 1-2: Contents of the status files (as of 2.6.30-rc7)
199 ..............................................................................
200 Field Content
201 Name filename of the executable
202 State state (R is running, S is sleeping, D is sleeping
203 in an uninterruptible wait, Z is zombie,
204 T is traced or stopped)
205 Tgid thread group ID
206 Pid process id
207 PPid process id of the parent process
208 TracerPid PID of process tracing this process (0 if not)
209 Uid Real, effective, saved set, and file system UIDs
210 Gid Real, effective, saved set, and file system GIDs
211 FDSize number of file descriptor slots currently allocated
212 Groups supplementary group list
213 VmPeak peak virtual memory size
214 VmSize total program size
215 VmLck locked memory size
216 VmHWM peak resident set size ("high water mark")
217 VmRSS size of memory portions
218 VmData size of data, stack, and text segments
219 VmStk size of data, stack, and text segments
220 VmExe size of text segment
221 VmLib size of shared library code
222 VmPTE size of page table entries
223 VmSwap size of swap usage (the number of referred swapents)
224 Threads number of threads
225 SigQ number of signals queued/max. number for queue
226 SigPnd bitmap of pending signals for the thread
227 ShdPnd bitmap of shared pending signals for the process
228 SigBlk bitmap of blocked signals
229 SigIgn bitmap of ignored signals
230 SigCgt bitmap of catched signals
231 CapInh bitmap of inheritable capabilities
232 CapPrm bitmap of permitted capabilities
233 CapEff bitmap of effective capabilities
234 CapBnd bitmap of capabilities bounding set
235 Cpus_allowed mask of CPUs on which this process may run
236 Cpus_allowed_list Same as previous, but in "list format"
237 Mems_allowed mask of memory nodes allowed to this process
238 Mems_allowed_list Same as previous, but in "list format"
239 voluntary_ctxt_switches number of voluntary context switches
240 nonvoluntary_ctxt_switches number of non voluntary context switches
241 ..............................................................................
243 Table 1-3: Contents of the statm files (as of 2.6.8-rc3)
244 ..............................................................................
245 Field Content
246 size total program size (pages) (same as VmSize in status)
247 resident size of memory portions (pages) (same as VmRSS in status)
248 shared number of pages that are shared (i.e. backed by a file)
249 trs number of pages that are 'code' (not including libs; broken,
250 includes data segment)
251 lrs number of pages of library (always 0 on 2.6)
252 drs number of pages of data/stack (including libs; broken,
253 includes library text)
254 dt number of dirty pages (always 0 on 2.6)
255 ..............................................................................
258 Table 1-4: Contents of the stat files (as of 2.6.30-rc7)
259 ..............................................................................
260 Field Content
261 pid process id
262 tcomm filename of the executable
263 state state (R is running, S is sleeping, D is sleeping in an
264 uninterruptible wait, Z is zombie, T is traced or stopped)
265 ppid process id of the parent process
266 pgrp pgrp of the process
267 sid session id
268 tty_nr tty the process uses
269 tty_pgrp pgrp of the tty
270 flags task flags
271 min_flt number of minor faults
272 cmin_flt number of minor faults with child's
273 maj_flt number of major faults
274 cmaj_flt number of major faults with child's
275 utime user mode jiffies
276 stime kernel mode jiffies
277 cutime user mode jiffies with child's
278 cstime kernel mode jiffies with child's
279 priority priority level
280 nice nice level
281 num_threads number of threads
282 it_real_value (obsolete, always 0)
283 start_time time the process started after system boot
284 vsize virtual memory size
285 rss resident set memory size
286 rsslim current limit in bytes on the rss
287 start_code address above which program text can run
288 end_code address below which program text can run
289 start_stack address of the start of the stack
290 esp current value of ESP
291 eip current value of EIP
292 pending bitmap of pending signals
293 blocked bitmap of blocked signals
294 sigign bitmap of ignored signals
295 sigcatch bitmap of catched signals
296 wchan address where process went to sleep
297 0 (place holder)
298 0 (place holder)
299 exit_signal signal to send to parent thread on exit
300 task_cpu which CPU the task is scheduled on
301 rt_priority realtime priority
302 policy scheduling policy (man sched_setscheduler)
303 blkio_ticks time spent waiting for block IO
304 gtime guest time of the task in jiffies
305 cgtime guest time of the task children in jiffies
306 ..............................................................................
308 The /proc/PID/maps file containing the currently mapped memory regions and
309 their access permissions.
311 The format is:
313 address perms offset dev inode pathname
315 08048000-08049000 r-xp 00000000 03:00 8312 /opt/test
316 08049000-0804a000 rw-p 00001000 03:00 8312 /opt/test
317 0804a000-0806b000 rw-p 00000000 00:00 0 [heap]
318 a7cb1000-a7cb2000 ---p 00000000 00:00 0
319 a7cb2000-a7eb2000 rw-p 00000000 00:00 0
320 a7eb2000-a7eb3000 ---p 00000000 00:00 0
321 a7eb3000-a7ed5000 rw-p 00000000 00:00 0
322 a7ed5000-a8008000 r-xp 00000000 03:00 4222 /lib/libc.so.6
323 a8008000-a800a000 r--p 00133000 03:00 4222 /lib/libc.so.6
324 a800a000-a800b000 rw-p 00135000 03:00 4222 /lib/libc.so.6
325 a800b000-a800e000 rw-p 00000000 00:00 0
326 a800e000-a8022000 r-xp 00000000 03:00 14462 /lib/libpthread.so.0
327 a8022000-a8023000 r--p 00013000 03:00 14462 /lib/libpthread.so.0
328 a8023000-a8024000 rw-p 00014000 03:00 14462 /lib/libpthread.so.0
329 a8024000-a8027000 rw-p 00000000 00:00 0
330 a8027000-a8043000 r-xp 00000000 03:00 8317 /lib/ld-linux.so.2
331 a8043000-a8044000 r--p 0001b000 03:00 8317 /lib/ld-linux.so.2
332 a8044000-a8045000 rw-p 0001c000 03:00 8317 /lib/ld-linux.so.2
333 aff35000-aff4a000 rw-p 00000000 00:00 0 [stack]
334 ffffe000-fffff000 r-xp 00000000 00:00 0 [vdso]
336 where "address" is the address space in the process that it occupies, "perms"
337 is a set of permissions:
339 r = read
340 w = write
341 x = execute
342 s = shared
343 p = private (copy on write)
345 "offset" is the offset into the mapping, "dev" is the device (major:minor), and
346 "inode" is the inode on that device. 0 indicates that no inode is associated
347 with the memory region, as the case would be with BSS (uninitialized data).
348 The "pathname" shows the name associated file for this mapping. If the mapping
349 is not associated with a file:
351 [heap] = the heap of the program
352 [stack] = the stack of the main process
353 [vdso] = the "virtual dynamic shared object",
354 the kernel system call handler
356 or if empty, the mapping is anonymous.
359 The /proc/PID/smaps is an extension based on maps, showing the memory
360 consumption for each of the process's mappings. For each of mappings there
361 is a series of lines such as the following:
363 08048000-080bc000 r-xp 00000000 03:02 13130 /bin/bash
364 Size: 1084 kB
365 Rss: 892 kB
366 Pss: 374 kB
367 Shared_Clean: 892 kB
368 Shared_Dirty: 0 kB
369 Private_Clean: 0 kB
370 Private_Dirty: 0 kB
371 Referenced: 892 kB
372 Swap: 0 kB
373 KernelPageSize: 4 kB
374 MMUPageSize: 4 kB
376 The first of these lines shows the same information as is displayed for the
377 mapping in /proc/PID/maps. The remaining lines show the size of the mapping,
378 the amount of the mapping that is currently resident in RAM, the "proportional
379 set size” (divide each shared page by the number of processes sharing it), the
380 number of clean and dirty shared pages in the mapping, and the number of clean
381 and dirty private pages in the mapping. The "Referenced" indicates the amount
382 of memory currently marked as referenced or accessed.
384 This file is only present if the CONFIG_MMU kernel configuration option is
385 enabled.
387 The /proc/PID/clear_refs is used to reset the PG_Referenced and ACCESSED/YOUNG
388 bits on both physical and virtual pages associated with a process.
389 To clear the bits for all the pages associated with the process
390 > echo 1 > /proc/PID/clear_refs
392 To clear the bits for the anonymous pages associated with the process
393 > echo 2 > /proc/PID/clear_refs
395 To clear the bits for the file mapped pages associated with the process
396 > echo 3 > /proc/PID/clear_refs
397 Any other value written to /proc/PID/clear_refs will have no effect.
400 1.2 Kernel data
401 ---------------
403 Similar to the process entries, the kernel data files give information about
404 the running kernel. The files used to obtain this information are contained in
405 /proc and are listed in Table 1-5. Not all of these will be present in your
406 system. It depends on the kernel configuration and the loaded modules, which
407 files are there, and which are missing.
409 Table 1-5: Kernel info in /proc
410 ..............................................................................
411 File Content
412 apm Advanced power management info
413 buddyinfo Kernel memory allocator information (see text) (2.5)
414 bus Directory containing bus specific information
415 cmdline Kernel command line
416 cpuinfo Info about the CPU
417 devices Available devices (block and character)
418 dma Used DMS channels
419 filesystems Supported filesystems
420 driver Various drivers grouped here, currently rtc (2.4)
421 execdomains Execdomains, related to security (2.4)
422 fb Frame Buffer devices (2.4)
423 fs File system parameters, currently nfs/exports (2.4)
424 ide Directory containing info about the IDE subsystem
425 interrupts Interrupt usage
426 iomem Memory map (2.4)
427 ioports I/O port usage
428 irq Masks for irq to cpu affinity (2.4)(smp?)
429 isapnp ISA PnP (Plug&Play) Info (2.4)
430 kcore Kernel core image (can be ELF or A.OUT(deprecated in 2.4))
431 kmsg Kernel messages
432 ksyms Kernel symbol table
433 loadavg Load average of last 1, 5 & 15 minutes
434 locks Kernel locks
435 meminfo Memory info
436 misc Miscellaneous
437 modules List of loaded modules
438 mounts Mounted filesystems
439 net Networking info (see text)
440 pagetypeinfo Additional page allocator information (see text) (2.5)
441 partitions Table of partitions known to the system
442 pci Deprecated info of PCI bus (new way -> /proc/bus/pci/,
443 decoupled by lspci (2.4)
444 rtc Real time clock
445 scsi SCSI info (see text)
446 slabinfo Slab pool info
447 softirqs softirq usage
448 stat Overall statistics
449 swaps Swap space utilization
450 sys See chapter 2
451 sysvipc Info of SysVIPC Resources (msg, sem, shm) (2.4)
452 tty Info of tty drivers
453 uptime System uptime
454 version Kernel version
455 video bttv info of video resources (2.4)
456 vmallocinfo Show vmalloced areas
457 ..............................................................................
459 You can, for example, check which interrupts are currently in use and what
460 they are used for by looking in the file /proc/interrupts:
462 > cat /proc/interrupts
463 CPU0
464 0: 8728810 XT-PIC timer
465 1: 895 XT-PIC keyboard
466 2: 0 XT-PIC cascade
467 3: 531695 XT-PIC aha152x
468 4: 2014133 XT-PIC serial
469 5: 44401 XT-PIC pcnet_cs
470 8: 2 XT-PIC rtc
471 11: 8 XT-PIC i82365
472 12: 182918 XT-PIC PS/2 Mouse
473 13: 1 XT-PIC fpu
474 14: 1232265 XT-PIC ide0
475 15: 7 XT-PIC ide1
476 NMI: 0
478 In 2.4.* a couple of lines where added to this file LOC & ERR (this time is the
479 output of a SMP machine):
481 > cat /proc/interrupts
483 CPU0 CPU1
484 0: 1243498 1214548 IO-APIC-edge timer
485 1: 8949 8958 IO-APIC-edge keyboard
486 2: 0 0 XT-PIC cascade
487 5: 11286 10161 IO-APIC-edge soundblaster
488 8: 1 0 IO-APIC-edge rtc
489 9: 27422 27407 IO-APIC-edge 3c503
490 12: 113645 113873 IO-APIC-edge PS/2 Mouse
491 13: 0 0 XT-PIC fpu
492 14: 22491 24012 IO-APIC-edge ide0
493 15: 2183 2415 IO-APIC-edge ide1
494 17: 30564 30414 IO-APIC-level eth0
495 18: 177 164 IO-APIC-level bttv
496 NMI: 2457961 2457959
497 LOC: 2457882 2457881
498 ERR: 2155
500 NMI is incremented in this case because every timer interrupt generates a NMI
501 (Non Maskable Interrupt) which is used by the NMI Watchdog to detect lockups.
503 LOC is the local interrupt counter of the internal APIC of every CPU.
505 ERR is incremented in the case of errors in the IO-APIC bus (the bus that
506 connects the CPUs in a SMP system. This means that an error has been detected,
507 the IO-APIC automatically retry the transmission, so it should not be a big
508 problem, but you should read the SMP-FAQ.
510 In 2.6.2* /proc/interrupts was expanded again. This time the goal was for
511 /proc/interrupts to display every IRQ vector in use by the system, not
512 just those considered 'most important'. The new vectors are:
514 THR -- interrupt raised when a machine check threshold counter
515 (typically counting ECC corrected errors of memory or cache) exceeds
516 a configurable threshold. Only available on some systems.
518 TRM -- a thermal event interrupt occurs when a temperature threshold
519 has been exceeded for the CPU. This interrupt may also be generated
520 when the temperature drops back to normal.
522 SPU -- a spurious interrupt is some interrupt that was raised then lowered
523 by some IO device before it could be fully processed by the APIC. Hence
524 the APIC sees the interrupt but does not know what device it came from.
525 For this case the APIC will generate the interrupt with a IRQ vector
526 of 0xff. This might also be generated by chipset bugs.
528 RES, CAL, TLB -- rescheduling, call and TLB flush interrupts are
529 sent from one CPU to another per the needs of the OS. Typically,
530 their statistics are used by kernel developers and interested users to
531 determine the occurrence of interrupts of the given type.
533 The above IRQ vectors are displayed only when relevent. For example,
534 the threshold vector does not exist on x86_64 platforms. Others are
535 suppressed when the system is a uniprocessor. As of this writing, only
536 i386 and x86_64 platforms support the new IRQ vector displays.
538 Of some interest is the introduction of the /proc/irq directory to 2.4.
539 It could be used to set IRQ to CPU affinity, this means that you can "hook" an
540 IRQ to only one CPU, or to exclude a CPU of handling IRQs. The contents of the
541 irq subdir is one subdir for each IRQ, and two files; default_smp_affinity and
542 prof_cpu_mask.
544 For example
545 > ls /proc/irq/
546 0 10 12 14 16 18 2 4 6 8 prof_cpu_mask
547 1 11 13 15 17 19 3 5 7 9 default_smp_affinity
548 > ls /proc/irq/0/
549 smp_affinity
551 smp_affinity is a bitmask, in which you can specify which CPUs can handle the
552 IRQ, you can set it by doing:
554 > echo 1 > /proc/irq/10/smp_affinity
556 This means that only the first CPU will handle the IRQ, but you can also echo
557 5 which means that only the first and fourth CPU can handle the IRQ.
559 The contents of each smp_affinity file is the same by default:
561 > cat /proc/irq/0/smp_affinity
562 ffffffff
564 The default_smp_affinity mask applies to all non-active IRQs, which are the
565 IRQs which have not yet been allocated/activated, and hence which lack a
566 /proc/irq/[0-9]* directory.
568 The node file on an SMP system shows the node to which the device using the IRQ
569 reports itself as being attached. This hardware locality information does not
570 include information about any possible driver locality preference.
572 prof_cpu_mask specifies which CPUs are to be profiled by the system wide
573 profiler. Default value is ffffffff (all cpus).
575 The way IRQs are routed is handled by the IO-APIC, and it's Round Robin
576 between all the CPUs which are allowed to handle it. As usual the kernel has
577 more info than you and does a better job than you, so the defaults are the
578 best choice for almost everyone.
580 There are three more important subdirectories in /proc: net, scsi, and sys.
581 The general rule is that the contents, or even the existence of these
582 directories, depend on your kernel configuration. If SCSI is not enabled, the
583 directory scsi may not exist. The same is true with the net, which is there
584 only when networking support is present in the running kernel.
586 The slabinfo file gives information about memory usage at the slab level.
587 Linux uses slab pools for memory management above page level in version 2.2.
588 Commonly used objects have their own slab pool (such as network buffers,
589 directory cache, and so on).
591 ..............................................................................
593 > cat /proc/buddyinfo
595 Node 0, zone DMA 0 4 5 4 4 3 ...
596 Node 0, zone Normal 1 0 0 1 101 8 ...
597 Node 0, zone HighMem 2 0 0 1 1 0 ...
599 External fragmentation is a problem under some workloads, and buddyinfo is a
600 useful tool for helping diagnose these problems. Buddyinfo will give you a
601 clue as to how big an area you can safely allocate, or why a previous
602 allocation failed.
604 Each column represents the number of pages of a certain order which are
605 available. In this case, there are 0 chunks of 2^0*PAGE_SIZE available in
606 ZONE_DMA, 4 chunks of 2^1*PAGE_SIZE in ZONE_DMA, 101 chunks of 2^4*PAGE_SIZE
607 available in ZONE_NORMAL, etc...
609 More information relevant to external fragmentation can be found in
610 pagetypeinfo.
612 > cat /proc/pagetypeinfo
613 Page block order: 9
614 Pages per block: 512
616 Free pages count per migrate type at order 0 1 2 3 4 5 6 7 8 9 10
617 Node 0, zone DMA, type Unmovable 0 0 0 1 1 1 1 1 1 1 0
618 Node 0, zone DMA, type Reclaimable 0 0 0 0 0 0 0 0 0 0 0
619 Node 0, zone DMA, type Movable 1 1 2 1 2 1 1 0 1 0 2
620 Node 0, zone DMA, type Reserve 0 0 0 0 0 0 0 0 0 1 0
621 Node 0, zone DMA, type Isolate 0 0 0 0 0 0 0 0 0 0 0
622 Node 0, zone DMA32, type Unmovable 103 54 77 1 1 1 11 8 7 1 9
623 Node 0, zone DMA32, type Reclaimable 0 0 2 1 0 0 0 0 1 0 0
624 Node 0, zone DMA32, type Movable 169 152 113 91 77 54 39 13 6 1 452
625 Node 0, zone DMA32, type Reserve 1 2 2 2 2 0 1 1 1 1 0
626 Node 0, zone DMA32, type Isolate 0 0 0 0 0 0 0 0 0 0 0
628 Number of blocks type Unmovable Reclaimable Movable Reserve Isolate
629 Node 0, zone DMA 2 0 5 1 0
630 Node 0, zone DMA32 41 6 967 2 0
632 Fragmentation avoidance in the kernel works by grouping pages of different
633 migrate types into the same contiguous regions of memory called page blocks.
634 A page block is typically the size of the default hugepage size e.g. 2MB on
635 X86-64. By keeping pages grouped based on their ability to move, the kernel
636 can reclaim pages within a page block to satisfy a high-order allocation.
638 The pagetypinfo begins with information on the size of a page block. It
639 then gives the same type of information as buddyinfo except broken down
640 by migrate-type and finishes with details on how many page blocks of each
641 type exist.
643 If min_free_kbytes has been tuned correctly (recommendations made by hugeadm
644 from libhugetlbfs http://sourceforge.net/projects/libhugetlbfs/), one can
645 make an estimate of the likely number of huge pages that can be allocated
646 at a given point in time. All the "Movable" blocks should be allocatable
647 unless memory has been mlock()'d. Some of the Reclaimable blocks should
648 also be allocatable although a lot of filesystem metadata may have to be
649 reclaimed to achieve this.
651 ..............................................................................
653 meminfo:
655 Provides information about distribution and utilization of memory. This
656 varies by architecture and compile options. The following is from a
657 16GB PIII, which has highmem enabled. You may not have all of these fields.
659 > cat /proc/meminfo
662 MemTotal: 16344972 kB
663 MemFree: 13634064 kB
664 Buffers: 3656 kB
665 Cached: 1195708 kB
666 SwapCached: 0 kB
667 Active: 891636 kB
668 Inactive: 1077224 kB
669 HighTotal: 15597528 kB
670 HighFree: 13629632 kB
671 LowTotal: 747444 kB
672 LowFree: 4432 kB
673 SwapTotal: 0 kB
674 SwapFree: 0 kB
675 Dirty: 968 kB
676 Writeback: 0 kB
677 AnonPages: 861800 kB
678 Mapped: 280372 kB
679 Slab: 284364 kB
680 SReclaimable: 159856 kB
681 SUnreclaim: 124508 kB
682 PageTables: 24448 kB
683 NFS_Unstable: 0 kB
684 Bounce: 0 kB
685 WritebackTmp: 0 kB
686 CommitLimit: 7669796 kB
687 Committed_AS: 100056 kB
688 VmallocTotal: 112216 kB
689 VmallocUsed: 428 kB
690 VmallocChunk: 111088 kB
692 MemTotal: Total usable ram (i.e. physical ram minus a few reserved
693 bits and the kernel binary code)
694 MemFree: The sum of LowFree+HighFree
695 Buffers: Relatively temporary storage for raw disk blocks
696 shouldn't get tremendously large (20MB or so)
697 Cached: in-memory cache for files read from the disk (the
698 pagecache). Doesn't include SwapCached
699 SwapCached: Memory that once was swapped out, is swapped back in but
700 still also is in the swapfile (if memory is needed it
701 doesn't need to be swapped out AGAIN because it is already
702 in the swapfile. This saves I/O)
703 Active: Memory that has been used more recently and usually not
704 reclaimed unless absolutely necessary.
705 Inactive: Memory which has been less recently used. It is more
706 eligible to be reclaimed for other purposes
707 HighTotal:
708 HighFree: Highmem is all memory above ~860MB of physical memory
709 Highmem areas are for use by userspace programs, or
710 for the pagecache. The kernel must use tricks to access
711 this memory, making it slower to access than lowmem.
712 LowTotal:
713 LowFree: Lowmem is memory which can be used for everything that
714 highmem can be used for, but it is also available for the
715 kernel's use for its own data structures. Among many
716 other things, it is where everything from the Slab is
717 allocated. Bad things happen when you're out of lowmem.
718 SwapTotal: total amount of swap space available
719 SwapFree: Memory which has been evicted from RAM, and is temporarily
720 on the disk
721 Dirty: Memory which is waiting to get written back to the disk
722 Writeback: Memory which is actively being written back to the disk
723 AnonPages: Non-file backed pages mapped into userspace page tables
724 Mapped: files which have been mmaped, such as libraries
725 Slab: in-kernel data structures cache
726 SReclaimable: Part of Slab, that might be reclaimed, such as caches
727 SUnreclaim: Part of Slab, that cannot be reclaimed on memory pressure
728 PageTables: amount of memory dedicated to the lowest level of page
729 tables.
730 NFS_Unstable: NFS pages sent to the server, but not yet committed to stable
731 storage
732 Bounce: Memory used for block device "bounce buffers"
733 WritebackTmp: Memory used by FUSE for temporary writeback buffers
734 CommitLimit: Based on the overcommit ratio ('vm.overcommit_ratio'),
735 this is the total amount of memory currently available to
736 be allocated on the system. This limit is only adhered to
737 if strict overcommit accounting is enabled (mode 2 in
738 'vm.overcommit_memory').
739 The CommitLimit is calculated with the following formula:
740 CommitLimit = ('vm.overcommit_ratio' * Physical RAM) + Swap
741 For example, on a system with 1G of physical RAM and 7G
742 of swap with a `vm.overcommit_ratio` of 30 it would
743 yield a CommitLimit of 7.3G.
744 For more details, see the memory overcommit documentation
745 in vm/overcommit-accounting.
746 Committed_AS: The amount of memory presently allocated on the system.
747 The committed memory is a sum of all of the memory which
748 has been allocated by processes, even if it has not been
749 "used" by them as of yet. A process which malloc()'s 1G
750 of memory, but only touches 300M of it will only show up
751 as using 300M of memory even if it has the address space
752 allocated for the entire 1G. This 1G is memory which has
753 been "committed" to by the VM and can be used at any time
754 by the allocating application. With strict overcommit
755 enabled on the system (mode 2 in 'vm.overcommit_memory'),
756 allocations which would exceed the CommitLimit (detailed
757 above) will not be permitted. This is useful if one needs
758 to guarantee that processes will not fail due to lack of
759 memory once that memory has been successfully allocated.
760 VmallocTotal: total size of vmalloc memory area
761 VmallocUsed: amount of vmalloc area which is used
762 VmallocChunk: largest contiguous block of vmalloc area which is free
764 ..............................................................................
766 vmallocinfo:
768 Provides information about vmalloced/vmaped areas. One line per area,
769 containing the virtual address range of the area, size in bytes,
770 caller information of the creator, and optional information depending
771 on the kind of area :
773 pages=nr number of pages
774 phys=addr if a physical address was specified
775 ioremap I/O mapping (ioremap() and friends)
776 vmalloc vmalloc() area
777 vmap vmap()ed pages
778 user VM_USERMAP area
779 vpages buffer for pages pointers was vmalloced (huge area)
780 N<node>=nr (Only on NUMA kernels)
781 Number of pages allocated on memory node <node>
783 > cat /proc/vmallocinfo
784 0xffffc20000000000-0xffffc20000201000 2101248 alloc_large_system_hash+0x204 ...
785 /0x2c0 pages=512 vmalloc N0=128 N1=128 N2=128 N3=128
786 0xffffc20000201000-0xffffc20000302000 1052672 alloc_large_system_hash+0x204 ...
787 /0x2c0 pages=256 vmalloc N0=64 N1=64 N2=64 N3=64
788 0xffffc20000302000-0xffffc20000304000 8192 acpi_tb_verify_table+0x21/0x4f...
789 phys=7fee8000 ioremap
790 0xffffc20000304000-0xffffc20000307000 12288 acpi_tb_verify_table+0x21/0x4f...
791 phys=7fee7000 ioremap
792 0xffffc2000031d000-0xffffc2000031f000 8192 init_vdso_vars+0x112/0x210
793 0xffffc2000031f000-0xffffc2000032b000 49152 cramfs_uncompress_init+0x2e ...
794 /0x80 pages=11 vmalloc N0=3 N1=3 N2=2 N3=3
795 0xffffc2000033a000-0xffffc2000033d000 12288 sys_swapon+0x640/0xac0 ...
796 pages=2 vmalloc N1=2
797 0xffffc20000347000-0xffffc2000034c000 20480 xt_alloc_table_info+0xfe ...
798 /0x130 [x_tables] pages=4 vmalloc N0=4
799 0xffffffffa0000000-0xffffffffa000f000 61440 sys_init_module+0xc27/0x1d00 ...
800 pages=14 vmalloc N2=14
801 0xffffffffa000f000-0xffffffffa0014000 20480 sys_init_module+0xc27/0x1d00 ...
802 pages=4 vmalloc N1=4
803 0xffffffffa0014000-0xffffffffa0017000 12288 sys_init_module+0xc27/0x1d00 ...
804 pages=2 vmalloc N1=2
805 0xffffffffa0017000-0xffffffffa0022000 45056 sys_init_module+0xc27/0x1d00 ...
806 pages=10 vmalloc N0=10
808 ..............................................................................
810 softirqs:
812 Provides counts of softirq handlers serviced since boot time, for each cpu.
814 > cat /proc/softirqs
815 CPU0 CPU1 CPU2 CPU3
816 HI: 0 0 0 0
817 TIMER: 27166 27120 27097 27034
818 NET_TX: 0 0 0 17
819 NET_RX: 42 0 0 39
820 BLOCK: 0 0 107 1121
821 TASKLET: 0 0 0 290
822 SCHED: 27035 26983 26971 26746
823 HRTIMER: 0 0 0 0
824 RCU: 1678 1769 2178 2250
827 1.3 IDE devices in /proc/ide
828 ----------------------------
830 The subdirectory /proc/ide contains information about all IDE devices of which
831 the kernel is aware. There is one subdirectory for each IDE controller, the
832 file drivers and a link for each IDE device, pointing to the device directory
833 in the controller specific subtree.
835 The file drivers contains general information about the drivers used for the
836 IDE devices:
838 > cat /proc/ide/drivers
839 ide-cdrom version 4.53
840 ide-disk version 1.08
842 More detailed information can be found in the controller specific
843 subdirectories. These are named ide0, ide1 and so on. Each of these
844 directories contains the files shown in table 1-6.
847 Table 1-6: IDE controller info in /proc/ide/ide?
848 ..............................................................................
849 File Content
850 channel IDE channel (0 or 1)
851 config Configuration (only for PCI/IDE bridge)
852 mate Mate name
853 model Type/Chipset of IDE controller
854 ..............................................................................
856 Each device connected to a controller has a separate subdirectory in the
857 controllers directory. The files listed in table 1-7 are contained in these
858 directories.
861 Table 1-7: IDE device information
862 ..............................................................................
863 File Content
864 cache The cache
865 capacity Capacity of the medium (in 512Byte blocks)
866 driver driver and version
867 geometry physical and logical geometry
868 identify device identify block
869 media media type
870 model device identifier
871 settings device setup
872 smart_thresholds IDE disk management thresholds
873 smart_values IDE disk management values
874 ..............................................................................
876 The most interesting file is settings. This file contains a nice overview of
877 the drive parameters:
879 # cat /proc/ide/ide0/hda/settings
880 name value min max mode
881 ---- ----- --- --- ----
882 bios_cyl 526 0 65535 rw
883 bios_head 255 0 255 rw
884 bios_sect 63 0 63 rw
885 breada_readahead 4 0 127 rw
886 bswap 0 0 1 r
887 file_readahead 72 0 2097151 rw
888 io_32bit 0 0 3 rw
889 keepsettings 0 0 1 rw
890 max_kb_per_request 122 1 127 rw
891 multcount 0 0 8 rw
892 nice1 1 0 1 rw
893 nowerr 0 0 1 rw
894 pio_mode write-only 0 255 w
895 slow 0 0 1 rw
896 unmaskirq 0 0 1 rw
897 using_dma 0 0 1 rw
900 1.4 Networking info in /proc/net
901 --------------------------------
903 The subdirectory /proc/net follows the usual pattern. Table 1-8 shows the
904 additional values you get for IP version 6 if you configure the kernel to
905 support this. Table 1-9 lists the files and their meaning.
908 Table 1-8: IPv6 info in /proc/net
909 ..............................................................................
910 File Content
911 udp6 UDP sockets (IPv6)
912 tcp6 TCP sockets (IPv6)
913 raw6 Raw device statistics (IPv6)
914 igmp6 IP multicast addresses, which this host joined (IPv6)
915 if_inet6 List of IPv6 interface addresses
916 ipv6_route Kernel routing table for IPv6
917 rt6_stats Global IPv6 routing tables statistics
918 sockstat6 Socket statistics (IPv6)
919 snmp6 Snmp data (IPv6)
920 ..............................................................................
923 Table 1-9: Network info in /proc/net
924 ..............................................................................
925 File Content
926 arp Kernel ARP table
927 dev network devices with statistics
928 dev_mcast the Layer2 multicast groups a device is listening too
929 (interface index, label, number of references, number of bound
930 addresses).
931 dev_stat network device status
932 ip_fwchains Firewall chain linkage
933 ip_fwnames Firewall chain names
934 ip_masq Directory containing the masquerading tables
935 ip_masquerade Major masquerading table
936 netstat Network statistics
937 raw raw device statistics
938 route Kernel routing table
939 rpc Directory containing rpc info
940 rt_cache Routing cache
941 snmp SNMP data
942 sockstat Socket statistics
943 tcp TCP sockets
944 tr_rif Token ring RIF routing table
945 udp UDP sockets
946 unix UNIX domain sockets
947 wireless Wireless interface data (Wavelan etc)
948 igmp IP multicast addresses, which this host joined
949 psched Global packet scheduler parameters.
950 netlink List of PF_NETLINK sockets
951 ip_mr_vifs List of multicast virtual interfaces
952 ip_mr_cache List of multicast routing cache
953 ..............................................................................
955 You can use this information to see which network devices are available in
956 your system and how much traffic was routed over those devices:
958 > cat /proc/net/dev
959 Inter-|Receive |[...
960 face |bytes packets errs drop fifo frame compressed multicast|[...
961 lo: 908188 5596 0 0 0 0 0 0 [...
962 ppp0:15475140 20721 410 0 0 410 0 0 [...
963 eth0: 614530 7085 0 0 0 0 0 1 [...
965 ...] Transmit
966 ...] bytes packets errs drop fifo colls carrier compressed
967 ...] 908188 5596 0 0 0 0 0 0
968 ...] 1375103 17405 0 0 0 0 0 0
969 ...] 1703981 5535 0 0 0 3 0 0
971 In addition, each Channel Bond interface has its own directory. For
972 example, the bond0 device will have a directory called /proc/net/bond0/.
973 It will contain information that is specific to that bond, such as the
974 current slaves of the bond, the link status of the slaves, and how
975 many times the slaves link has failed.
977 1.5 SCSI info
978 -------------
980 If you have a SCSI host adapter in your system, you'll find a subdirectory
981 named after the driver for this adapter in /proc/scsi. You'll also see a list
982 of all recognized SCSI devices in /proc/scsi:
984 >cat /proc/scsi/scsi
985 Attached devices:
986 Host: scsi0 Channel: 00 Id: 00 Lun: 00
987 Vendor: IBM Model: DGHS09U Rev: 03E0
988 Type: Direct-Access ANSI SCSI revision: 03
989 Host: scsi0 Channel: 00 Id: 06 Lun: 00
990 Vendor: PIONEER Model: CD-ROM DR-U06S Rev: 1.04
991 Type: CD-ROM ANSI SCSI revision: 02
994 The directory named after the driver has one file for each adapter found in
995 the system. These files contain information about the controller, including
996 the used IRQ and the IO address range. The amount of information shown is
997 dependent on the adapter you use. The example shows the output for an Adaptec
998 AHA-2940 SCSI adapter:
1000 > cat /proc/scsi/aic7xxx/0
1002 Adaptec AIC7xxx driver version: 5.1.19/3.2.4
1003 Compile Options:
1004 TCQ Enabled By Default : Disabled
1005 AIC7XXX_PROC_STATS : Disabled
1006 AIC7XXX_RESET_DELAY : 5
1007 Adapter Configuration:
1008 SCSI Adapter: Adaptec AHA-294X Ultra SCSI host adapter
1009 Ultra Wide Controller
1010 PCI MMAPed I/O Base: 0xeb001000
1011 Adapter SEEPROM Config: SEEPROM found and used.
1012 Adaptec SCSI BIOS: Enabled
1013 IRQ: 10
1014 SCBs: Active 0, Max Active 2,
1015 Allocated 15, HW 16, Page 255
1016 Interrupts: 160328
1017 BIOS Control Word: 0x18b6
1018 Adapter Control Word: 0x005b
1019 Extended Translation: Enabled
1020 Disconnect Enable Flags: 0xffff
1021 Ultra Enable Flags: 0x0001
1022 Tag Queue Enable Flags: 0x0000
1023 Ordered Queue Tag Flags: 0x0000
1024 Default Tag Queue Depth: 8
1025 Tagged Queue By Device array for aic7xxx host instance 0:
1026 {255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255}
1027 Actual queue depth per device for aic7xxx host instance 0:
1028 {1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1}
1029 Statistics:
1030 (scsi0:0:0:0)
1031 Device using Wide/Sync transfers at 40.0 MByte/sec, offset 8
1032 Transinfo settings: current(12/8/1/0), goal(12/8/1/0), user(12/15/1/0)
1033 Total transfers 160151 (74577 reads and 85574 writes)
1034 (scsi0:0:6:0)
1035 Device using Narrow/Sync transfers at 5.0 MByte/sec, offset 15
1036 Transinfo settings: current(50/15/0/0), goal(50/15/0/0), user(50/15/0/0)
1037 Total transfers 0 (0 reads and 0 writes)
1040 1.6 Parallel port info in /proc/parport
1041 ---------------------------------------
1043 The directory /proc/parport contains information about the parallel ports of
1044 your system. It has one subdirectory for each port, named after the port
1045 number (0,1,2,...).
1047 These directories contain the four files shown in Table 1-10.
1050 Table 1-10: Files in /proc/parport
1051 ..............................................................................
1052 File Content
1053 autoprobe Any IEEE-1284 device ID information that has been acquired.
1054 devices list of the device drivers using that port. A + will appear by the
1055 name of the device currently using the port (it might not appear
1056 against any).
1057 hardware Parallel port's base address, IRQ line and DMA channel.
1058 irq IRQ that parport is using for that port. This is in a separate
1059 file to allow you to alter it by writing a new value in (IRQ
1060 number or none).
1061 ..............................................................................
1063 1.7 TTY info in /proc/tty
1064 -------------------------
1066 Information about the available and actually used tty's can be found in the
1067 directory /proc/tty.You'll find entries for drivers and line disciplines in
1068 this directory, as shown in Table 1-11.
1071 Table 1-11: Files in /proc/tty
1072 ..............................................................................
1073 File Content
1074 drivers list of drivers and their usage
1075 ldiscs registered line disciplines
1076 driver/serial usage statistic and status of single tty lines
1077 ..............................................................................
1079 To see which tty's are currently in use, you can simply look into the file
1080 /proc/tty/drivers:
1082 > cat /proc/tty/drivers
1083 pty_slave /dev/pts 136 0-255 pty:slave
1084 pty_master /dev/ptm 128 0-255 pty:master
1085 pty_slave /dev/ttyp 3 0-255 pty:slave
1086 pty_master /dev/pty 2 0-255 pty:master
1087 serial /dev/cua 5 64-67 serial:callout
1088 serial /dev/ttyS 4 64-67 serial
1089 /dev/tty0 /dev/tty0 4 0 system:vtmaster
1090 /dev/ptmx /dev/ptmx 5 2 system
1091 /dev/console /dev/console 5 1 system:console
1092 /dev/tty /dev/tty 5 0 system:/dev/tty
1093 unknown /dev/tty 4 1-63 console
1096 1.8 Miscellaneous kernel statistics in /proc/stat
1097 -------------------------------------------------
1099 Various pieces of information about kernel activity are available in the
1100 /proc/stat file. All of the numbers reported in this file are aggregates
1101 since the system first booted. For a quick look, simply cat the file:
1103 > cat /proc/stat
1104 cpu 2255 34 2290 22625563 6290 127 456 0 0
1105 cpu0 1132 34 1441 11311718 3675 127 438 0 0
1106 cpu1 1123 0 849 11313845 2614 0 18 0 0
1107 intr 114930548 113199788 3 0 5 263 0 4 [... lots more numbers ...]
1108 ctxt 1990473
1109 btime 1062191376
1110 processes 2915
1111 procs_running 1
1112 procs_blocked 0
1113 softirq 183433 0 21755 12 39 1137 231 21459 2263
1115 The very first "cpu" line aggregates the numbers in all of the other "cpuN"
1116 lines. These numbers identify the amount of time the CPU has spent performing
1117 different kinds of work. Time units are in USER_HZ (typically hundredths of a
1118 second). The meanings of the columns are as follows, from left to right:
1120 - user: normal processes executing in user mode
1121 - nice: niced processes executing in user mode
1122 - system: processes executing in kernel mode
1123 - idle: twiddling thumbs
1124 - iowait: waiting for I/O to complete
1125 - irq: servicing interrupts
1126 - softirq: servicing softirqs
1127 - steal: involuntary wait
1128 - guest: running a normal guest
1129 - guest_nice: running a niced guest
1131 The "intr" line gives counts of interrupts serviced since boot time, for each
1132 of the possible system interrupts. The first column is the total of all
1133 interrupts serviced; each subsequent column is the total for that particular
1134 interrupt.
1136 The "ctxt" line gives the total number of context switches across all CPUs.
1138 The "btime" line gives the time at which the system booted, in seconds since
1139 the Unix epoch.
1141 The "processes" line gives the number of processes and threads created, which
1142 includes (but is not limited to) those created by calls to the fork() and
1143 clone() system calls.
1145 The "procs_running" line gives the total number of threads that are
1146 running or ready to run (i.e., the total number of runnable threads).
1148 The "procs_blocked" line gives the number of processes currently blocked,
1149 waiting for I/O to complete.
1151 The "softirq" line gives counts of softirqs serviced since boot time, for each
1152 of the possible system softirqs. The first column is the total of all
1153 softirqs serviced; each subsequent column is the total for that particular
1154 softirq.
1157 1.9 Ext4 file system parameters
1158 ------------------------------
1160 Information about mounted ext4 file systems can be found in
1161 /proc/fs/ext4. Each mounted filesystem will have a directory in
1162 /proc/fs/ext4 based on its device name (i.e., /proc/fs/ext4/hdc or
1163 /proc/fs/ext4/dm-0). The files in each per-device directory are shown
1164 in Table 1-12, below.
1166 Table 1-12: Files in /proc/fs/ext4/<devname>
1167 ..............................................................................
1168 File Content
1169 mb_groups details of multiblock allocator buddy cache of free blocks
1170 ..............................................................................
1173 ------------------------------------------------------------------------------
1174 Summary
1175 ------------------------------------------------------------------------------
1176 The /proc file system serves information about the running system. It not only
1177 allows access to process data but also allows you to request the kernel status
1178 by reading files in the hierarchy.
1180 The directory structure of /proc reflects the types of information and makes
1181 it easy, if not obvious, where to look for specific data.
1182 ------------------------------------------------------------------------------
1184 ------------------------------------------------------------------------------
1185 CHAPTER 2: MODIFYING SYSTEM PARAMETERS
1186 ------------------------------------------------------------------------------
1188 ------------------------------------------------------------------------------
1189 In This Chapter
1190 ------------------------------------------------------------------------------
1191 * Modifying kernel parameters by writing into files found in /proc/sys
1192 * Exploring the files which modify certain parameters
1193 * Review of the /proc/sys file tree
1194 ------------------------------------------------------------------------------
1197 A very interesting part of /proc is the directory /proc/sys. This is not only
1198 a source of information, it also allows you to change parameters within the
1199 kernel. Be very careful when attempting this. You can optimize your system,
1200 but you can also cause it to crash. Never alter kernel parameters on a
1201 production system. Set up a development machine and test to make sure that
1202 everything works the way you want it to. You may have no alternative but to
1203 reboot the machine once an error has been made.
1205 To change a value, simply echo the new value into the file. An example is
1206 given below in the section on the file system data. You need to be root to do
1207 this. You can create your own boot script to perform this every time your
1208 system boots.
1210 The files in /proc/sys can be used to fine tune and monitor miscellaneous and
1211 general things in the operation of the Linux kernel. Since some of the files
1212 can inadvertently disrupt your system, it is advisable to read both
1213 documentation and source before actually making adjustments. In any case, be
1214 very careful when writing to any of these files. The entries in /proc may
1215 change slightly between the 2.1.* and the 2.2 kernel, so if there is any doubt
1216 review the kernel documentation in the directory /usr/src/linux/Documentation.
1217 This chapter is heavily based on the documentation included in the pre 2.2
1218 kernels, and became part of it in version 2.2.1 of the Linux kernel.
1220 Please see: Documentation/sysctls/ directory for descriptions of these
1221 entries.
1223 ------------------------------------------------------------------------------
1224 Summary
1225 ------------------------------------------------------------------------------
1226 Certain aspects of kernel behavior can be modified at runtime, without the
1227 need to recompile the kernel, or even to reboot the system. The files in the
1228 /proc/sys tree can not only be read, but also modified. You can use the echo
1229 command to write value into these files, thereby changing the default settings
1230 of the kernel.
1231 ------------------------------------------------------------------------------
1233 ------------------------------------------------------------------------------
1234 CHAPTER 3: PER-PROCESS PARAMETERS
1235 ------------------------------------------------------------------------------
1237 3.1 /proc/<pid>/oom_adj - Adjust the oom-killer score
1238 ------------------------------------------------------
1240 This file can be used to adjust the score used to select which processes
1241 should be killed in an out-of-memory situation. Giving it a high score will
1242 increase the likelihood of this process being killed by the oom-killer. Valid
1243 values are in the range -16 to +15, plus the special value -17, which disables
1244 oom-killing altogether for this process.
1246 The process to be killed in an out-of-memory situation is selected among all others
1247 based on its badness score. This value equals the original memory size of the process
1248 and is then updated according to its CPU time (utime + stime) and the
1249 run time (uptime - start time). The longer it runs the smaller is the score.
1250 Badness score is divided by the square root of the CPU time and then by
1251 the double square root of the run time.
1253 Swapped out tasks are killed first. Half of each child's memory size is added to
1254 the parent's score if they do not share the same memory. Thus forking servers
1255 are the prime candidates to be killed. Having only one 'hungry' child will make
1256 parent less preferable than the child.
1258 /proc/<pid>/oom_score shows process' current badness score.
1260 The following heuristics are then applied:
1261 * if the task was reniced, its score doubles
1262 * superuser or direct hardware access tasks (CAP_SYS_ADMIN, CAP_SYS_RESOURCE
1263 or CAP_SYS_RAWIO) have their score divided by 4
1264 * if oom condition happened in one cpuset and checked process does not belong
1265 to it, its score is divided by 8
1266 * the resulting score is multiplied by two to the power of oom_adj, i.e.
1267 points <<= oom_adj when it is positive and
1268 points >>= -(oom_adj) otherwise
1270 The task with the highest badness score is then selected and its children
1271 are killed, process itself will be killed in an OOM situation when it does
1272 not have children or some of them disabled oom like described above.
1274 3.2 /proc/<pid>/oom_score - Display current oom-killer score
1275 -------------------------------------------------------------
1277 This file can be used to check the current score used by the oom-killer is for
1278 any given <pid>. Use it together with /proc/<pid>/oom_adj to tune which
1279 process should be killed in an out-of-memory situation.
1282 3.3 /proc/<pid>/io - Display the IO accounting fields
1283 -------------------------------------------------------
1285 This file contains IO statistics for each running process
1287 Example
1288 -------
1290 test:/tmp # dd if=/dev/zero of=/tmp/test.dat &
1291 [1] 3828
1293 test:/tmp # cat /proc/3828/io
1294 rchar: 323934931
1295 wchar: 323929600
1296 syscr: 632687
1297 syscw: 632675
1298 read_bytes: 0
1299 write_bytes: 323932160
1300 cancelled_write_bytes: 0
1303 Description
1304 -----------
1306 rchar
1307 -----
1309 I/O counter: chars read
1310 The number of bytes which this task has caused to be read from storage. This
1311 is simply the sum of bytes which this process passed to read() and pread().
1312 It includes things like tty IO and it is unaffected by whether or not actual
1313 physical disk IO was required (the read might have been satisfied from
1314 pagecache)
1317 wchar
1318 -----
1320 I/O counter: chars written
1321 The number of bytes which this task has caused, or shall cause to be written
1322 to disk. Similar caveats apply here as with rchar.
1325 syscr
1326 -----
1328 I/O counter: read syscalls
1329 Attempt to count the number of read I/O operations, i.e. syscalls like read()
1330 and pread().
1333 syscw
1334 -----
1336 I/O counter: write syscalls
1337 Attempt to count the number of write I/O operations, i.e. syscalls like
1338 write() and pwrite().
1341 read_bytes
1342 ----------
1344 I/O counter: bytes read
1345 Attempt to count the number of bytes which this process really did cause to
1346 be fetched from the storage layer. Done at the submit_bio() level, so it is
1347 accurate for block-backed filesystems. <please add status regarding NFS and
1348 CIFS at a later time>
1351 write_bytes
1352 -----------
1354 I/O counter: bytes written
1355 Attempt to count the number of bytes which this process caused to be sent to
1356 the storage layer. This is done at page-dirtying time.
1359 cancelled_write_bytes
1360 ---------------------
1362 The big inaccuracy here is truncate. If a process writes 1MB to a file and
1363 then deletes the file, it will in fact perform no writeout. But it will have
1364 been accounted as having caused 1MB of write.
1365 In other words: The number of bytes which this process caused to not happen,
1366 by truncating pagecache. A task can cause "negative" IO too. If this task
1367 truncates some dirty pagecache, some IO which another task has been accounted
1368 for (in its write_bytes) will not be happening. We _could_ just subtract that
1369 from the truncating task's write_bytes, but there is information loss in doing
1370 that.
1373 Note
1374 ----
1376 At its current implementation state, this is a bit racy on 32-bit machines: if
1377 process A reads process B's /proc/pid/io while process B is updating one of
1378 those 64-bit counters, process A could see an intermediate result.
1381 More information about this can be found within the taskstats documentation in
1382 Documentation/accounting.
1384 3.4 /proc/<pid>/coredump_filter - Core dump filtering settings
1385 ---------------------------------------------------------------
1386 When a process is dumped, all anonymous memory is written to a core file as
1387 long as the size of the core file isn't limited. But sometimes we don't want
1388 to dump some memory segments, for example, huge shared memory. Conversely,
1389 sometimes we want to save file-backed memory segments into a core file, not
1390 only the individual files.
1392 /proc/<pid>/coredump_filter allows you to customize which memory segments
1393 will be dumped when the <pid> process is dumped. coredump_filter is a bitmask
1394 of memory types. If a bit of the bitmask is set, memory segments of the
1395 corresponding memory type are dumped, otherwise they are not dumped.
1397 The following 7 memory types are supported:
1398 - (bit 0) anonymous private memory
1399 - (bit 1) anonymous shared memory
1400 - (bit 2) file-backed private memory
1401 - (bit 3) file-backed shared memory
1402 - (bit 4) ELF header pages in file-backed private memory areas (it is
1403 effective only if the bit 2 is cleared)
1404 - (bit 5) hugetlb private memory
1405 - (bit 6) hugetlb shared memory
1407 Note that MMIO pages such as frame buffer are never dumped and vDSO pages
1408 are always dumped regardless of the bitmask status.
1410 Note bit 0-4 doesn't effect any hugetlb memory. hugetlb memory are only
1411 effected by bit 5-6.
1413 Default value of coredump_filter is 0x23; this means all anonymous memory
1414 segments and hugetlb private memory are dumped.
1416 If you don't want to dump all shared memory segments attached to pid 1234,
1417 write 0x21 to the process's proc file.
1419 $ echo 0x21 > /proc/1234/coredump_filter
1421 When a new process is created, the process inherits the bitmask status from its
1422 parent. It is useful to set up coredump_filter before the program runs.
1423 For example:
1425 $ echo 0x7 > /proc/self/coredump_filter
1426 $ ./some_program
1428 3.5 /proc/<pid>/mountinfo - Information about mounts
1429 --------------------------------------------------------
1431 This file contains lines of the form:
1433 36 35 98:0 /mnt1 /mnt2 rw,noatime master:1 - ext3 /dev/root rw,errors=continue
1434 (1)(2)(3) (4) (5) (6) (7) (8) (9) (10) (11)
1436 (1) mount ID: unique identifier of the mount (may be reused after umount)
1437 (2) parent ID: ID of parent (or of self for the top of the mount tree)
1438 (3) major:minor: value of st_dev for files on filesystem
1439 (4) root: root of the mount within the filesystem
1440 (5) mount point: mount point relative to the process's root
1441 (6) mount options: per mount options
1442 (7) optional fields: zero or more fields of the form "tag[:value]"
1443 (8) separator: marks the end of the optional fields
1444 (9) filesystem type: name of filesystem of the form "type[.subtype]"
1445 (10) mount source: filesystem specific information or "none"
1446 (11) super options: per super block options
1448 Parsers should ignore all unrecognised optional fields. Currently the
1449 possible optional fields are:
1451 shared:X mount is shared in peer group X
1452 master:X mount is slave to peer group X
1453 propagate_from:X mount is slave and receives propagation from peer group X (*)
1454 unbindable mount is unbindable
1456 (*) X is the closest dominant peer group under the process's root. If
1457 X is the immediate master of the mount, or if there's no dominant peer
1458 group under the same root, then only the "master:X" field is present
1459 and not the "propagate_from:X" field.
1461 For more information on mount propagation see:
1463 Documentation/filesystems/sharedsubtree.txt
1466 3.6 /proc/<pid>/comm & /proc/<pid>/task/<tid>/comm
1467 --------------------------------------------------------
1468 These files provide a method to access a tasks comm value. It also allows for
1469 a task to set its own or one of its thread siblings comm value. The comm value
1470 is limited in size compared to the cmdline value, so writing anything longer
1471 then the kernel's TASK_COMM_LEN (currently 16 chars) will result in a truncated
1472 comm value.