1Documentation for /proc/sys/kernel/* kernel version 2.2.10 2 (c) 1998, 1999, Rik van Riel <riel@nl.linux.org> 3 (c) 2009, Shen Feng<shen@cn.fujitsu.com> 4 5For general info and legal blurb, please look in README. 6 7============================================================== 8 9This file contains documentation for the sysctl files in 10/proc/sys/kernel/ and is valid for Linux kernel version 2.2. 11 12The files in this directory can be used to tune and monitor 13miscellaneous and general things in the operation of the Linux 14kernel. Since some of the files _can_ be used to screw up your 15system, it is advisable to read both documentation and source 16before actually making adjustments. 17 18Currently, these files might (depending on your configuration) 19show up in /proc/sys/kernel: 20 21- acct 22- acpi_video_flags 23- auto_msgmni 24- bootloader_type [ X86 only ] 25- bootloader_version [ X86 only ] 26- callhome [ S390 only ] 27- cap_last_cap 28- core_pattern 29- core_pipe_limit 30- core_uses_pid 31- ctrl-alt-del 32- dmesg_restrict 33- domainname 34- hostname 35- hotplug 36- kptr_restrict 37- kstack_depth_to_print [ X86 only ] 38- l2cr [ PPC only ] 39- modprobe ==> Documentation/debugging-modules.txt 40- modules_disabled 41- msg_next_id [ sysv ipc ] 42- msgmax 43- msgmnb 44- msgmni 45- nmi_watchdog 46- osrelease 47- ostype 48- overflowgid 49- overflowuid 50- panic 51- panic_on_oops 52- panic_on_stackoverflow 53- panic_on_unrecovered_nmi 54- perf_event_paranoid 55- pid_max 56- powersave-nap [ PPC only ] 57- printk 58- printk_delay 59- printk_ratelimit 60- printk_ratelimit_burst 61- randomize_va_space 62- real-root-dev ==> Documentation/initrd.txt 63- reboot-cmd [ SPARC only ] 64- rtsig-max 65- rtsig-nr 66- sem 67- sem_next_id [ sysv ipc ] 68- sg-big-buff [ generic SCSI device (sg) ] 69- shm_next_id [ sysv ipc ] 70- shm_rmid_forced 71- shmall 72- shmmax [ sysv ipc ] 73- shmmni 74- softlockup_thresh 75- stop-a [ SPARC only ] 76- sysrq ==> Documentation/sysrq.txt 77- tainted 78- threads-max 79- unknown_nmi_panic 80- version 81 82============================================================== 83 84acct: 85 86highwater lowwater frequency 87 88If BSD-style process accounting is enabled these values control 89its behaviour. If free space on filesystem where the log lives 90goes below <lowwater>% accounting suspends. If free space gets 91above <highwater>% accounting resumes. <Frequency> determines 92how often do we check the amount of free space (value is in 93seconds). Default: 944 2 30 95That is, suspend accounting if there left <= 2% free; resume it 96if we got >=4%; consider information about amount of free space 97valid for 30 seconds. 98 99============================================================== 100 101acpi_video_flags: 102 103flags 104 105See Doc*/kernel/power/video.txt, it allows mode of video boot to be 106set during run time. 107 108============================================================== 109 110auto_msgmni: 111 112Enables/Disables automatic recomputing of msgmni upon memory add/remove 113or upon ipc namespace creation/removal (see the msgmni description 114above). Echoing "1" into this file enables msgmni automatic recomputing. 115Echoing "0" turns it off. auto_msgmni default value is 1. 116 117 118============================================================== 119 120bootloader_type: 121 122x86 bootloader identification 123 124This gives the bootloader type number as indicated by the bootloader, 125shifted left by 4, and OR'd with the low four bits of the bootloader 126version. The reason for this encoding is that this used to match the 127type_of_loader field in the kernel header; the encoding is kept for 128backwards compatibility. That is, if the full bootloader type number 129is 0x15 and the full version number is 0x234, this file will contain 130the value 340 = 0x154. 131 132See the type_of_loader and ext_loader_type fields in 133Documentation/x86/boot.txt for additional information. 134 135============================================================== 136 137bootloader_version: 138 139x86 bootloader version 140 141The complete bootloader version number. In the example above, this 142file will contain the value 564 = 0x234. 143 144See the type_of_loader and ext_loader_ver fields in 145Documentation/x86/boot.txt for additional information. 146 147============================================================== 148 149callhome: 150 151Controls the kernel's callhome behavior in case of a kernel panic. 152 153The s390 hardware allows an operating system to send a notification 154to a service organization (callhome) in case of an operating system panic. 155 156When the value in this file is 0 (which is the default behavior) 157nothing happens in case of a kernel panic. If this value is set to "1" 158the complete kernel oops message is send to the IBM customer service 159organization in case the mainframe the Linux operating system is running 160on has a service contract with IBM. 161 162============================================================== 163 164cap_last_cap 165 166Highest valid capability of the running kernel. Exports 167CAP_LAST_CAP from the kernel. 168 169============================================================== 170 171core_pattern: 172 173core_pattern is used to specify a core dumpfile pattern name. 174. max length 128 characters; default value is "core" 175. core_pattern is used as a pattern template for the output filename; 176 certain string patterns (beginning with '%') are substituted with 177 their actual values. 178. backward compatibility with core_uses_pid: 179 If core_pattern does not include "%p" (default does not) 180 and core_uses_pid is set, then .PID will be appended to 181 the filename. 182. corename format specifiers: 183 %<NUL> '%' is dropped 184 %% output one '%' 185 %p pid 186 %u uid 187 %g gid 188 %d dump mode, matches PR_SET_DUMPABLE and 189 /proc/sys/fs/suid_dumpable 190 %s signal number 191 %t UNIX time of dump 192 %h hostname 193 %e executable filename (may be shortened) 194 %E executable path 195 %<OTHER> both are dropped 196. If the first character of the pattern is a '|', the kernel will treat 197 the rest of the pattern as a command to run. The core dump will be 198 written to the standard input of that program instead of to a file. 199 200============================================================== 201 202core_pipe_limit: 203 204This sysctl is only applicable when core_pattern is configured to pipe 205core files to a user space helper (when the first character of 206core_pattern is a '|', see above). When collecting cores via a pipe 207to an application, it is occasionally useful for the collecting 208application to gather data about the crashing process from its 209/proc/pid directory. In order to do this safely, the kernel must wait 210for the collecting process to exit, so as not to remove the crashing 211processes proc files prematurely. This in turn creates the 212possibility that a misbehaving userspace collecting process can block 213the reaping of a crashed process simply by never exiting. This sysctl 214defends against that. It defines how many concurrent crashing 215processes may be piped to user space applications in parallel. If 216this value is exceeded, then those crashing processes above that value 217are noted via the kernel log and their cores are skipped. 0 is a 218special value, indicating that unlimited processes may be captured in 219parallel, but that no waiting will take place (i.e. the collecting 220process is not guaranteed access to /proc/<crashing pid>/). This 221value defaults to 0. 222 223============================================================== 224 225core_uses_pid: 226 227The default coredump filename is "core". By setting 228core_uses_pid to 1, the coredump filename becomes core.PID. 229If core_pattern does not include "%p" (default does not) 230and core_uses_pid is set, then .PID will be appended to 231the filename. 232 233============================================================== 234 235ctrl-alt-del: 236 237When the value in this file is 0, ctrl-alt-del is trapped and 238sent to the init(1) program to handle a graceful restart. 239When, however, the value is > 0, Linux's reaction to a Vulcan 240Nerve Pinch (tm) will be an immediate reboot, without even 241syncing its dirty buffers. 242 243Note: when a program (like dosemu) has the keyboard in 'raw' 244mode, the ctrl-alt-del is intercepted by the program before it 245ever reaches the kernel tty layer, and it's up to the program 246to decide what to do with it. 247 248============================================================== 249 250dmesg_restrict: 251 252This toggle indicates whether unprivileged users are prevented 253from using dmesg(8) to view messages from the kernel's log buffer. 254When dmesg_restrict is set to (0) there are no restrictions. When 255dmesg_restrict is set set to (1), users must have CAP_SYSLOG to use 256dmesg(8). 257 258The kernel config option CONFIG_SECURITY_DMESG_RESTRICT sets the 259default value of dmesg_restrict. 260 261============================================================== 262 263domainname & hostname: 264 265These files can be used to set the NIS/YP domainname and the 266hostname of your box in exactly the same way as the commands 267domainname and hostname, i.e.: 268# echo "darkstar" > /proc/sys/kernel/hostname 269# echo "mydomain" > /proc/sys/kernel/domainname 270has the same effect as 271# hostname "darkstar" 272# domainname "mydomain" 273 274Note, however, that the classic darkstar.frop.org has the 275hostname "darkstar" and DNS (Internet Domain Name Server) 276domainname "frop.org", not to be confused with the NIS (Network 277Information Service) or YP (Yellow Pages) domainname. These two 278domain names are in general different. For a detailed discussion 279see the hostname(1) man page. 280 281============================================================== 282 283hotplug: 284 285Path for the hotplug policy agent. 286Default value is "/sbin/hotplug". 287 288============================================================== 289 290kptr_restrict: 291 292This toggle indicates whether restrictions are placed on 293exposing kernel addresses via /proc and other interfaces. When 294kptr_restrict is set to (0), there are no restrictions. When 295kptr_restrict is set to (1), the default, kernel pointers 296printed using the %pK format specifier will be replaced with 0's 297unless the user has CAP_SYSLOG. When kptr_restrict is set to 298(2), kernel pointers printed using %pK will be replaced with 0's 299regardless of privileges. 300 301============================================================== 302 303kstack_depth_to_print: (X86 only) 304 305Controls the number of words to print when dumping the raw 306kernel stack. 307 308============================================================== 309 310l2cr: (PPC only) 311 312This flag controls the L2 cache of G3 processor boards. If 3130, the cache is disabled. Enabled if nonzero. 314 315============================================================== 316 317modules_disabled: 318 319A toggle value indicating if modules are allowed to be loaded 320in an otherwise modular kernel. This toggle defaults to off 321(0), but can be set true (1). Once true, modules can be 322neither loaded nor unloaded, and the toggle cannot be set back 323to false. 324 325============================================================== 326 327msg_next_id, sem_next_id, and shm_next_id: 328 329These three toggles allows to specify desired id for next allocated IPC 330object: message, semaphore or shared memory respectively. 331 332By default they are equal to -1, which means generic allocation logic. 333Possible values to set are in range {0..INT_MAX}. 334 335Notes: 3361) kernel doesn't guarantee, that new object will have desired id. So, 337it's up to userspace, how to handle an object with "wrong" id. 3382) Toggle with non-default value will be set back to -1 by kernel after 339successful IPC object allocation. 340 341============================================================== 342 343nmi_watchdog: 344 345Enables/Disables the NMI watchdog on x86 systems. When the value is 346non-zero the NMI watchdog is enabled and will continuously test all 347online cpus to determine whether or not they are still functioning 348properly. Currently, passing "nmi_watchdog=" parameter at boot time is 349required for this function to work. 350 351If LAPIC NMI watchdog method is in use (nmi_watchdog=2 kernel 352parameter), the NMI watchdog shares registers with oprofile. By 353disabling the NMI watchdog, oprofile may have more registers to 354utilize. 355 356============================================================== 357 358osrelease, ostype & version: 359 360# cat osrelease 3612.1.88 362# cat ostype 363Linux 364# cat version 365#5 Wed Feb 25 21:49:24 MET 1998 366 367The files osrelease and ostype should be clear enough. Version 368needs a little more clarification however. The '#5' means that 369this is the fifth kernel built from this source base and the 370date behind it indicates the time the kernel was built. 371The only way to tune these values is to rebuild the kernel :-) 372 373============================================================== 374 375overflowgid & overflowuid: 376 377if your architecture did not always support 32-bit UIDs (i.e. arm, 378i386, m68k, sh, and sparc32), a fixed UID and GID will be returned to 379applications that use the old 16-bit UID/GID system calls, if the 380actual UID or GID would exceed 65535. 381 382These sysctls allow you to change the value of the fixed UID and GID. 383The default is 65534. 384 385============================================================== 386 387panic: 388 389The value in this file represents the number of seconds the kernel 390waits before rebooting on a panic. When you use the software watchdog, 391the recommended setting is 60. 392 393============================================================== 394 395panic_on_oops: 396 397Controls the kernel's behaviour when an oops or BUG is encountered. 398 3990: try to continue operation 400 4011: panic immediately. If the `panic' sysctl is also non-zero then the 402 machine will be rebooted. 403 404============================================================== 405 406panic_on_stackoverflow: 407 408Controls the kernel's behavior when detecting the overflows of 409kernel, IRQ and exception stacks except a user stack. 410This file shows up if CONFIG_DEBUG_STACKOVERFLOW is enabled. 411 4120: try to continue operation. 413 4141: panic immediately. 415 416============================================================== 417 418panic_on_unrecovered_nmi: 419 420The default Linux behaviour on an NMI of either memory or unknown is 421to continue operation. For many environments such as scientific 422computing it is preferable that the box is taken out and the error 423dealt with than an uncorrected parity/ECC error get propagated. 424 425A small number of systems do generate NMI's for bizarre random reasons 426such as power management so the default is off. That sysctl works like 427the existing panic controls already in that directory. 428 429============================================================== 430 431perf_event_paranoid: 432 433Controls use of the performance events system by unprivileged 434users (without CAP_SYS_ADMIN). The default value is 3 if 435CONFIG_SECURITY_PERF_EVENTS_RESTRICT is set, or 1 otherwise. 436 437 -1: Allow use of (almost) all events by all users 438>=0: Disallow raw tracepoint access by users without CAP_IOC_LOCK 439>=1: Disallow CPU event access by users without CAP_SYS_ADMIN 440>=2: Disallow kernel profiling by users without CAP_SYS_ADMIN 441>=3: Disallow all event access by users without CAP_SYS_ADMIN 442 443============================================================== 444 445pid_max: 446 447PID allocation wrap value. When the kernel's next PID value 448reaches this value, it wraps back to a minimum PID value. 449PIDs of value pid_max or larger are not allocated. 450 451============================================================== 452 453ns_last_pid: 454 455The last pid allocated in the current (the one task using this sysctl 456lives in) pid namespace. When selecting a pid for a next task on fork 457kernel tries to allocate a number starting from this one. 458 459============================================================== 460 461powersave-nap: (PPC only) 462 463If set, Linux-PPC will use the 'nap' mode of powersaving, 464otherwise the 'doze' mode will be used. 465 466============================================================== 467 468printk: 469 470The four values in printk denote: console_loglevel, 471default_message_loglevel, minimum_console_loglevel and 472default_console_loglevel respectively. 473 474These values influence printk() behavior when printing or 475logging error messages. See 'man 2 syslog' for more info on 476the different loglevels. 477 478- console_loglevel: messages with a higher priority than 479 this will be printed to the console 480- default_message_loglevel: messages without an explicit priority 481 will be printed with this priority 482- minimum_console_loglevel: minimum (highest) value to which 483 console_loglevel can be set 484- default_console_loglevel: default value for console_loglevel 485 486============================================================== 487 488printk_delay: 489 490Delay each printk message in printk_delay milliseconds 491 492Value from 0 - 10000 is allowed. 493 494============================================================== 495 496printk_ratelimit: 497 498Some warning messages are rate limited. printk_ratelimit specifies 499the minimum length of time between these messages (in jiffies), by 500default we allow one every 5 seconds. 501 502A value of 0 will disable rate limiting. 503 504============================================================== 505 506printk_ratelimit_burst: 507 508While long term we enforce one message per printk_ratelimit 509seconds, we do allow a burst of messages to pass through. 510printk_ratelimit_burst specifies the number of messages we can 511send before ratelimiting kicks in. 512 513============================================================== 514 515randomize_va_space: 516 517This option can be used to select the type of process address 518space randomization that is used in the system, for architectures 519that support this feature. 520 5210 - Turn the process address space randomization off. This is the 522 default for architectures that do not support this feature anyways, 523 and kernels that are booted with the "norandmaps" parameter. 524 5251 - Make the addresses of mmap base, stack and VDSO page randomized. 526 This, among other things, implies that shared libraries will be 527 loaded to random addresses. Also for PIE-linked binaries, the 528 location of code start is randomized. This is the default if the 529 CONFIG_COMPAT_BRK option is enabled. 530 5312 - Additionally enable heap randomization. This is the default if 532 CONFIG_COMPAT_BRK is disabled. 533 534 There are a few legacy applications out there (such as some ancient 535 versions of libc.so.5 from 1996) that assume that brk area starts 536 just after the end of the code+bss. These applications break when 537 start of the brk area is randomized. There are however no known 538 non-legacy applications that would be broken this way, so for most 539 systems it is safe to choose full randomization. 540 541 Systems with ancient and/or broken binaries should be configured 542 with CONFIG_COMPAT_BRK enabled, which excludes the heap from process 543 address space randomization. 544 545============================================================== 546 547reboot-cmd: (Sparc only) 548 549??? This seems to be a way to give an argument to the Sparc 550ROM/Flash boot loader. Maybe to tell it what to do after 551rebooting. ??? 552 553============================================================== 554 555rtsig-max & rtsig-nr: 556 557The file rtsig-max can be used to tune the maximum number 558of POSIX realtime (queued) signals that can be outstanding 559in the system. 560 561rtsig-nr shows the number of RT signals currently queued. 562 563============================================================== 564 565sg-big-buff: 566 567This file shows the size of the generic SCSI (sg) buffer. 568You can't tune it just yet, but you could change it on 569compile time by editing include/scsi/sg.h and changing 570the value of SG_BIG_BUFF. 571 572There shouldn't be any reason to change this value. If 573you can come up with one, you probably know what you 574are doing anyway :) 575 576============================================================== 577 578shmall: 579 580This parameter sets the total amount of shared memory pages that 581can be used system wide. Hence, SHMALL should always be at least 582ceil(shmmax/PAGE_SIZE). 583 584If you are not sure what the default PAGE_SIZE is on your Linux 585system, you can run the following command: 586 587# getconf PAGE_SIZE 588 589============================================================== 590 591shmmax: 592 593This value can be used to query and set the run time limit 594on the maximum shared memory segment size that can be created. 595Shared memory segments up to 1Gb are now supported in the 596kernel. This value defaults to SHMMAX. 597 598============================================================== 599 600shm_rmid_forced: 601 602Linux lets you set resource limits, including how much memory one 603process can consume, via setrlimit(2). Unfortunately, shared memory 604segments are allowed to exist without association with any process, and 605thus might not be counted against any resource limits. If enabled, 606shared memory segments are automatically destroyed when their attach 607count becomes zero after a detach or a process termination. It will 608also destroy segments that were created, but never attached to, on exit 609from the process. The only use left for IPC_RMID is to immediately 610destroy an unattached segment. Of course, this breaks the way things are 611defined, so some applications might stop working. Note that this 612feature will do you no good unless you also configure your resource 613limits (in particular, RLIMIT_AS and RLIMIT_NPROC). Most systems don't 614need this. 615 616Note that if you change this from 0 to 1, already created segments 617without users and with a dead originative process will be destroyed. 618 619============================================================== 620 621softlockup_thresh: 622 623This value can be used to lower the softlockup tolerance threshold. The 624default threshold is 60 seconds. If a cpu is locked up for 60 seconds, 625the kernel complains. Valid values are 1-60 seconds. Setting this 626tunable to zero will disable the softlockup detection altogether. 627 628============================================================== 629 630tainted: 631 632Non-zero if the kernel has been tainted. Numeric values, which 633can be ORed together: 634 635 1 - A module with a non-GPL license has been loaded, this 636 includes modules with no license. 637 Set by modutils >= 2.4.9 and module-init-tools. 638 2 - A module was force loaded by insmod -f. 639 Set by modutils >= 2.4.9 and module-init-tools. 640 4 - Unsafe SMP processors: SMP with CPUs not designed for SMP. 641 8 - A module was forcibly unloaded from the system by rmmod -f. 642 16 - A hardware machine check error occurred on the system. 643 32 - A bad page was discovered on the system. 644 64 - The user has asked that the system be marked "tainted". This 645 could be because they are running software that directly modifies 646 the hardware, or for other reasons. 647 128 - The system has died. 648 256 - The ACPI DSDT has been overridden with one supplied by the user 649 instead of using the one provided by the hardware. 650 512 - A kernel warning has occurred. 6511024 - A module from drivers/staging was loaded. 6522048 - The system is working around a severe firmware bug. 6534096 - An out-of-tree module has been loaded. 654 655============================================================== 656 657unknown_nmi_panic: 658 659The value in this file affects behavior of handling NMI. When the 660value is non-zero, unknown NMI is trapped and then panic occurs. At 661that time, kernel debugging information is displayed on console. 662 663NMI switch that most IA32 servers have fires unknown NMI up, for 664example. If a system hangs up, try pressing the NMI switch. 665