Commit | Line | Data |
---|---|---|
1da177e4 LT |
1 | Documentation for /proc/sys/kernel/* kernel version 2.2.10 |
2 | (c) 1998, 1999, Rik van Riel <riel@nl.linux.org> | |
760df93e | 3 | (c) 2009, Shen Feng<shen@cn.fujitsu.com> |
1da177e4 LT |
4 | |
5 | For general info and legal blurb, please look in README. | |
6 | ||
7 | ============================================================== | |
8 | ||
9 | This file contains documentation for the sysctl files in | |
10 | /proc/sys/kernel/ and is valid for Linux kernel version 2.2. | |
11 | ||
12 | The files in this directory can be used to tune and monitor | |
13 | miscellaneous and general things in the operation of the Linux | |
14 | kernel. Since some of the files _can_ be used to screw up your | |
15 | system, it is advisable to read both documentation and source | |
16 | before actually making adjustments. | |
17 | ||
18 | Currently, these files might (depending on your configuration) | |
19 | show up in /proc/sys/kernel: | |
c255d844 | 20 | - acpi_video_flags |
1da177e4 | 21 | - acct |
c114728a | 22 | - callhome [ S390 only ] |
760df93e | 23 | - auto_msgmni |
1da177e4 | 24 | - core_pattern |
a293980c | 25 | - core_pipe_limit |
1da177e4 LT |
26 | - core_uses_pid |
27 | - ctrl-alt-del | |
28 | - dentry-state | |
29 | - domainname | |
30 | - hostname | |
31 | - hotplug | |
32 | - java-appletviewer [ binfmt_java, obsolete ] | |
33 | - java-interpreter [ binfmt_java, obsolete ] | |
0741f4d2 | 34 | - kstack_depth_to_print [ X86 only ] |
1da177e4 | 35 | - l2cr [ PPC only ] |
ac76cff2 | 36 | - modprobe ==> Documentation/debugging-modules.txt |
3d43321b | 37 | - modules_disabled |
1da177e4 LT |
38 | - msgmax |
39 | - msgmnb | |
40 | - msgmni | |
760df93e | 41 | - nmi_watchdog |
1da177e4 LT |
42 | - osrelease |
43 | - ostype | |
44 | - overflowgid | |
45 | - overflowuid | |
46 | - panic | |
47 | - pid_max | |
48 | - powersave-nap [ PPC only ] | |
760df93e | 49 | - panic_on_unrecovered_nmi |
1da177e4 | 50 | - printk |
1ec7fd50 | 51 | - randomize_va_space |
1da177e4 LT |
52 | - real-root-dev ==> Documentation/initrd.txt |
53 | - reboot-cmd [ SPARC only ] | |
54 | - rtsig-max | |
55 | - rtsig-nr | |
56 | - sem | |
57 | - sg-big-buff [ generic SCSI device (sg) ] | |
58 | - shmall | |
59 | - shmmax [ sysv ipc ] | |
60 | - shmmni | |
61 | - stop-a [ SPARC only ] | |
62 | - sysrq ==> Documentation/sysrq.txt | |
63 | - tainted | |
64 | - threads-max | |
760df93e | 65 | - unknown_nmi_panic |
1da177e4 LT |
66 | - version |
67 | ||
68 | ============================================================== | |
69 | ||
c255d844 PM |
70 | acpi_video_flags: |
71 | ||
72 | flags | |
73 | ||
74 | See Doc*/kernel/power/video.txt, it allows mode of video boot to be | |
75 | set during run time. | |
76 | ||
77 | ============================================================== | |
78 | ||
1da177e4 LT |
79 | acct: |
80 | ||
81 | highwater lowwater frequency | |
82 | ||
83 | If BSD-style process accounting is enabled these values control | |
84 | its behaviour. If free space on filesystem where the log lives | |
85 | goes below <lowwater>% accounting suspends. If free space gets | |
86 | above <highwater>% accounting resumes. <Frequency> determines | |
87 | how often do we check the amount of free space (value is in | |
88 | seconds). Default: | |
89 | 4 2 30 | |
90 | That is, suspend accounting if there left <= 2% free; resume it | |
91 | if we got >=4%; consider information about amount of free space | |
92 | valid for 30 seconds. | |
93 | ||
94 | ============================================================== | |
95 | ||
c114728a HJP |
96 | callhome: |
97 | ||
98 | Controls the kernel's callhome behavior in case of a kernel panic. | |
99 | ||
100 | The s390 hardware allows an operating system to send a notification | |
101 | to a service organization (callhome) in case of an operating system panic. | |
102 | ||
103 | When the value in this file is 0 (which is the default behavior) | |
104 | nothing happens in case of a kernel panic. If this value is set to "1" | |
105 | the complete kernel oops message is send to the IBM customer service | |
106 | organization in case the mainframe the Linux operating system is running | |
107 | on has a service contract with IBM. | |
108 | ||
109 | ============================================================== | |
110 | ||
1da177e4 LT |
111 | core_pattern: |
112 | ||
113 | core_pattern is used to specify a core dumpfile pattern name. | |
cd081041 | 114 | . max length 128 characters; default value is "core" |
1da177e4 LT |
115 | . core_pattern is used as a pattern template for the output filename; |
116 | certain string patterns (beginning with '%') are substituted with | |
117 | their actual values. | |
118 | . backward compatibility with core_uses_pid: | |
119 | If core_pattern does not include "%p" (default does not) | |
120 | and core_uses_pid is set, then .PID will be appended to | |
121 | the filename. | |
122 | . corename format specifiers: | |
123 | %<NUL> '%' is dropped | |
124 | %% output one '%' | |
125 | %p pid | |
126 | %u uid | |
127 | %g gid | |
128 | %s signal number | |
129 | %t UNIX time of dump | |
130 | %h hostname | |
131 | %e executable filename | |
132 | %<OTHER> both are dropped | |
cd081041 MU |
133 | . If the first character of the pattern is a '|', the kernel will treat |
134 | the rest of the pattern as a command to run. The core dump will be | |
135 | written to the standard input of that program instead of to a file. | |
1da177e4 LT |
136 | |
137 | ============================================================== | |
138 | ||
a293980c NH |
139 | core_pipe_limit: |
140 | ||
141 | This sysctl is only applicable when core_pattern is configured to pipe core | |
142 | files to user space helper a (when the first character of core_pattern is a '|', | |
143 | see above). When collecting cores via a pipe to an application, it is | |
144 | occasionally usefull for the collecting application to gather data about the | |
145 | crashing process from its /proc/pid directory. In order to do this safely, the | |
146 | kernel must wait for the collecting process to exit, so as not to remove the | |
147 | crashing processes proc files prematurely. This in turn creates the possibility | |
148 | that a misbehaving userspace collecting process can block the reaping of a | |
149 | crashed process simply by never exiting. This sysctl defends against that. It | |
150 | defines how many concurrent crashing processes may be piped to user space | |
151 | applications in parallel. If this value is exceeded, then those crashing | |
152 | processes above that value are noted via the kernel log and their cores are | |
153 | skipped. 0 is a special value, indicating that unlimited processes may be | |
154 | captured in parallel, but that no waiting will take place (i.e. the collecting | |
155 | process is not guaranteed access to /proc/<crahing pid>/). This value defaults | |
156 | to 0. | |
157 | ||
158 | ============================================================== | |
159 | ||
1da177e4 LT |
160 | core_uses_pid: |
161 | ||
162 | The default coredump filename is "core". By setting | |
163 | core_uses_pid to 1, the coredump filename becomes core.PID. | |
164 | If core_pattern does not include "%p" (default does not) | |
165 | and core_uses_pid is set, then .PID will be appended to | |
166 | the filename. | |
167 | ||
168 | ============================================================== | |
169 | ||
170 | ctrl-alt-del: | |
171 | ||
172 | When the value in this file is 0, ctrl-alt-del is trapped and | |
173 | sent to the init(1) program to handle a graceful restart. | |
174 | When, however, the value is > 0, Linux's reaction to a Vulcan | |
175 | Nerve Pinch (tm) will be an immediate reboot, without even | |
176 | syncing its dirty buffers. | |
177 | ||
178 | Note: when a program (like dosemu) has the keyboard in 'raw' | |
179 | mode, the ctrl-alt-del is intercepted by the program before it | |
180 | ever reaches the kernel tty layer, and it's up to the program | |
181 | to decide what to do with it. | |
182 | ||
183 | ============================================================== | |
184 | ||
185 | domainname & hostname: | |
186 | ||
187 | These files can be used to set the NIS/YP domainname and the | |
188 | hostname of your box in exactly the same way as the commands | |
189 | domainname and hostname, i.e.: | |
190 | # echo "darkstar" > /proc/sys/kernel/hostname | |
191 | # echo "mydomain" > /proc/sys/kernel/domainname | |
192 | has the same effect as | |
193 | # hostname "darkstar" | |
194 | # domainname "mydomain" | |
195 | ||
196 | Note, however, that the classic darkstar.frop.org has the | |
197 | hostname "darkstar" and DNS (Internet Domain Name Server) | |
198 | domainname "frop.org", not to be confused with the NIS (Network | |
199 | Information Service) or YP (Yellow Pages) domainname. These two | |
200 | domain names are in general different. For a detailed discussion | |
201 | see the hostname(1) man page. | |
202 | ||
203 | ============================================================== | |
204 | ||
205 | hotplug: | |
206 | ||
207 | Path for the hotplug policy agent. | |
208 | Default value is "/sbin/hotplug". | |
209 | ||
210 | ============================================================== | |
211 | ||
212 | l2cr: (PPC only) | |
213 | ||
214 | This flag controls the L2 cache of G3 processor boards. If | |
215 | 0, the cache is disabled. Enabled if nonzero. | |
216 | ||
217 | ============================================================== | |
218 | ||
0741f4d2 CE |
219 | kstack_depth_to_print: (X86 only) |
220 | ||
221 | Controls the number of words to print when dumping the raw | |
222 | kernel stack. | |
223 | ||
224 | ============================================================== | |
225 | ||
3d43321b KC |
226 | modules_disabled: |
227 | ||
228 | A toggle value indicating if modules are allowed to be loaded | |
229 | in an otherwise modular kernel. This toggle defaults to off | |
230 | (0), but can be set true (1). Once true, modules can be | |
231 | neither loaded nor unloaded, and the toggle cannot be set back | |
232 | to false. | |
233 | ||
234 | ============================================================== | |
235 | ||
1da177e4 LT |
236 | osrelease, ostype & version: |
237 | ||
238 | # cat osrelease | |
239 | 2.1.88 | |
240 | # cat ostype | |
241 | Linux | |
242 | # cat version | |
243 | #5 Wed Feb 25 21:49:24 MET 1998 | |
244 | ||
245 | The files osrelease and ostype should be clear enough. Version | |
246 | needs a little more clarification however. The '#5' means that | |
247 | this is the fifth kernel built from this source base and the | |
248 | date behind it indicates the time the kernel was built. | |
249 | The only way to tune these values is to rebuild the kernel :-) | |
250 | ||
251 | ============================================================== | |
252 | ||
253 | overflowgid & overflowuid: | |
254 | ||
255 | if your architecture did not always support 32-bit UIDs (i.e. arm, i386, | |
256 | m68k, sh, and sparc32), a fixed UID and GID will be returned to | |
257 | applications that use the old 16-bit UID/GID system calls, if the actual | |
258 | UID or GID would exceed 65535. | |
259 | ||
260 | These sysctls allow you to change the value of the fixed UID and GID. | |
261 | The default is 65534. | |
262 | ||
263 | ============================================================== | |
264 | ||
265 | panic: | |
266 | ||
267 | The value in this file represents the number of seconds the | |
268 | kernel waits before rebooting on a panic. When you use the | |
269 | software watchdog, the recommended setting is 60. | |
270 | ||
271 | ============================================================== | |
272 | ||
273 | panic_on_oops: | |
274 | ||
275 | Controls the kernel's behaviour when an oops or BUG is encountered. | |
276 | ||
277 | 0: try to continue operation | |
278 | ||
a982ac06 | 279 | 1: panic immediately. If the `panic' sysctl is also non-zero then the |
8b23d04d | 280 | machine will be rebooted. |
1da177e4 LT |
281 | |
282 | ============================================================== | |
283 | ||
284 | pid_max: | |
285 | ||
beb7dd86 | 286 | PID allocation wrap value. When the kernel's next PID value |
1da177e4 LT |
287 | reaches this value, it wraps back to a minimum PID value. |
288 | PIDs of value pid_max or larger are not allocated. | |
289 | ||
290 | ============================================================== | |
291 | ||
292 | powersave-nap: (PPC only) | |
293 | ||
294 | If set, Linux-PPC will use the 'nap' mode of powersaving, | |
295 | otherwise the 'doze' mode will be used. | |
296 | ||
297 | ============================================================== | |
298 | ||
299 | printk: | |
300 | ||
301 | The four values in printk denote: console_loglevel, | |
302 | default_message_loglevel, minimum_console_loglevel and | |
303 | default_console_loglevel respectively. | |
304 | ||
305 | These values influence printk() behavior when printing or | |
306 | logging error messages. See 'man 2 syslog' for more info on | |
307 | the different loglevels. | |
308 | ||
309 | - console_loglevel: messages with a higher priority than | |
310 | this will be printed to the console | |
311 | - default_message_level: messages without an explicit priority | |
312 | will be printed with this priority | |
313 | - minimum_console_loglevel: minimum (highest) value to which | |
314 | console_loglevel can be set | |
315 | - default_console_loglevel: default value for console_loglevel | |
316 | ||
317 | ============================================================== | |
318 | ||
319 | printk_ratelimit: | |
320 | ||
321 | Some warning messages are rate limited. printk_ratelimit specifies | |
322 | the minimum length of time between these messages (in jiffies), by | |
323 | default we allow one every 5 seconds. | |
324 | ||
325 | A value of 0 will disable rate limiting. | |
326 | ||
327 | ============================================================== | |
328 | ||
329 | printk_ratelimit_burst: | |
330 | ||
331 | While long term we enforce one message per printk_ratelimit | |
332 | seconds, we do allow a burst of messages to pass through. | |
333 | printk_ratelimit_burst specifies the number of messages we can | |
334 | send before ratelimiting kicks in. | |
335 | ||
336 | ============================================================== | |
337 | ||
af91322e DY |
338 | printk_delay: |
339 | ||
340 | Delay each printk message in printk_delay milliseconds | |
341 | ||
342 | Value from 0 - 10000 is allowed. | |
343 | ||
344 | ============================================================== | |
345 | ||
1ec7fd50 JK |
346 | randomize-va-space: |
347 | ||
348 | This option can be used to select the type of process address | |
349 | space randomization that is used in the system, for architectures | |
350 | that support this feature. | |
351 | ||
b7f5ab6f HS |
352 | 0 - Turn the process address space randomization off. This is the |
353 | default for architectures that do not support this feature anyways, | |
354 | and kernels that are booted with the "norandmaps" parameter. | |
1ec7fd50 JK |
355 | |
356 | 1 - Make the addresses of mmap base, stack and VDSO page randomized. | |
357 | This, among other things, implies that shared libraries will be | |
b7f5ab6f HS |
358 | loaded to random addresses. Also for PIE-linked binaries, the |
359 | location of code start is randomized. This is the default if the | |
360 | CONFIG_COMPAT_BRK option is enabled. | |
1ec7fd50 | 361 | |
b7f5ab6f HS |
362 | 2 - Additionally enable heap randomization. This is the default if |
363 | CONFIG_COMPAT_BRK is disabled. | |
364 | ||
365 | There are a few legacy applications out there (such as some ancient | |
1ec7fd50 | 366 | versions of libc.so.5 from 1996) that assume that brk area starts |
b7f5ab6f HS |
367 | just after the end of the code+bss. These applications break when |
368 | start of the brk area is randomized. There are however no known | |
1ec7fd50 | 369 | non-legacy applications that would be broken this way, so for most |
b7f5ab6f HS |
370 | systems it is safe to choose full randomization. |
371 | ||
372 | Systems with ancient and/or broken binaries should be configured | |
373 | with CONFIG_COMPAT_BRK enabled, which excludes the heap from process | |
374 | address space randomization. | |
1ec7fd50 JK |
375 | |
376 | ============================================================== | |
377 | ||
1da177e4 LT |
378 | reboot-cmd: (Sparc only) |
379 | ||
380 | ??? This seems to be a way to give an argument to the Sparc | |
381 | ROM/Flash boot loader. Maybe to tell it what to do after | |
382 | rebooting. ??? | |
383 | ||
384 | ============================================================== | |
385 | ||
386 | rtsig-max & rtsig-nr: | |
387 | ||
388 | The file rtsig-max can be used to tune the maximum number | |
389 | of POSIX realtime (queued) signals that can be outstanding | |
390 | in the system. | |
391 | ||
392 | rtsig-nr shows the number of RT signals currently queued. | |
393 | ||
394 | ============================================================== | |
395 | ||
396 | sg-big-buff: | |
397 | ||
398 | This file shows the size of the generic SCSI (sg) buffer. | |
399 | You can't tune it just yet, but you could change it on | |
400 | compile time by editing include/scsi/sg.h and changing | |
401 | the value of SG_BIG_BUFF. | |
402 | ||
403 | There shouldn't be any reason to change this value. If | |
404 | you can come up with one, you probably know what you | |
405 | are doing anyway :) | |
406 | ||
407 | ============================================================== | |
408 | ||
409 | shmmax: | |
410 | ||
411 | This value can be used to query and set the run time limit | |
412 | on the maximum shared memory segment size that can be created. | |
413 | Shared memory segments up to 1Gb are now supported in the | |
414 | kernel. This value defaults to SHMMAX. | |
415 | ||
416 | ============================================================== | |
417 | ||
c4f3b63f RT |
418 | softlockup_thresh: |
419 | ||
b4d19cc8 AM |
420 | This value can be used to lower the softlockup tolerance threshold. The |
421 | default threshold is 60 seconds. If a cpu is locked up for 60 seconds, | |
422 | the kernel complains. Valid values are 1-60 seconds. Setting this | |
423 | tunable to zero will disable the softlockup detection altogether. | |
c4f3b63f RT |
424 | |
425 | ============================================================== | |
426 | ||
1da177e4 LT |
427 | tainted: |
428 | ||
429 | Non-zero if the kernel has been tainted. Numeric values, which | |
430 | can be ORed together: | |
431 | ||
bb20698d GKH |
432 | 1 - A module with a non-GPL license has been loaded, this |
433 | includes modules with no license. | |
434 | Set by modutils >= 2.4.9 and module-init-tools. | |
435 | 2 - A module was force loaded by insmod -f. | |
436 | Set by modutils >= 2.4.9 and module-init-tools. | |
437 | 4 - Unsafe SMP processors: SMP with CPUs not designed for SMP. | |
438 | 8 - A module was forcibly unloaded from the system by rmmod -f. | |
439 | 16 - A hardware machine check error occurred on the system. | |
440 | 32 - A bad page was discovered on the system. | |
441 | 64 - The user has asked that the system be marked "tainted". This | |
442 | could be because they are running software that directly modifies | |
443 | the hardware, or for other reasons. | |
444 | 128 - The system has died. | |
445 | 256 - The ACPI DSDT has been overridden with one supplied by the user | |
446 | instead of using the one provided by the hardware. | |
447 | 512 - A kernel warning has occurred. | |
448 | 1024 - A module from drivers/staging was loaded. | |
1da177e4 | 449 | |
760df93e SF |
450 | ============================================================== |
451 | ||
452 | auto_msgmni: | |
453 | ||
454 | Enables/Disables automatic recomputing of msgmni upon memory add/remove or | |
455 | upon ipc namespace creation/removal (see the msgmni description above). | |
456 | Echoing "1" into this file enables msgmni automatic recomputing. | |
457 | Echoing "0" turns it off. | |
458 | auto_msgmni default value is 1. | |
459 | ||
460 | ============================================================== | |
461 | ||
462 | nmi_watchdog: | |
463 | ||
464 | Enables/Disables the NMI watchdog on x86 systems. When the value is non-zero | |
465 | the NMI watchdog is enabled and will continuously test all online cpus to | |
466 | determine whether or not they are still functioning properly. Currently, | |
467 | passing "nmi_watchdog=" parameter at boot time is required for this function | |
468 | to work. | |
469 | ||
470 | If LAPIC NMI watchdog method is in use (nmi_watchdog=2 kernel parameter), the | |
471 | NMI watchdog shares registers with oprofile. By disabling the NMI watchdog, | |
472 | oprofile may have more registers to utilize. | |
473 | ||
474 | ============================================================== | |
475 | ||
476 | unknown_nmi_panic: | |
477 | ||
478 | The value in this file affects behavior of handling NMI. When the value is | |
479 | non-zero, unknown NMI is trapped and then panic occurs. At that time, kernel | |
480 | debugging information is displayed on console. | |
481 | ||
482 | NMI switch that most IA32 servers have fires unknown NMI up, for example. | |
483 | If a system hangs up, try pressing the NMI switch. | |
484 | ||
485 | ============================================================== | |
486 | ||
487 | panic_on_unrecovered_nmi: | |
488 | ||
489 | The default Linux behaviour on an NMI of either memory or unknown is to continue | |
490 | operation. For many environments such as scientific computing it is preferable | |
491 | that the box is taken out and the error dealt with than an uncorrected | |
492 | parity/ECC error get propogated. | |
493 | ||
494 | A small number of systems do generate NMI's for bizarre random reasons such as | |
495 | power management so the default is off. That sysctl works like the existing | |
496 | panic controls already in that directory. | |
497 |