linux/Documentation/sysrq.txt
<<
>>
Prefs
   1Linux Magic System Request Key Hacks
   2Documentation for sysrq.c
   3
   4*  What is the magic SysRq key?
   5~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
   6It is a 'magical' key combo you can hit which the kernel will respond to
   7regardless of whatever else it is doing, unless it is completely locked up.
   8
   9*  How do I enable the magic SysRq key?
  10~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
  11You need to say "yes" to 'Magic SysRq key (CONFIG_MAGIC_SYSRQ)' when
  12configuring the kernel. When running a kernel with SysRq compiled in,
  13/proc/sys/kernel/sysrq controls the functions allowed to be invoked via
  14the SysRq key. The default value in this file is set by the
  15CONFIG_MAGIC_SYSRQ_DEFAULT_ENABLE config symbol, which itself defaults
  16to 1. Here is the list of possible values in /proc/sys/kernel/sysrq:
  17   0 - disable sysrq completely
  18   1 - enable all functions of sysrq
  19  >1 - bitmask of allowed sysrq functions (see below for detailed function
  20       description):
  21          2 =   0x2 - enable control of console logging level
  22          4 =   0x4 - enable control of keyboard (SAK, unraw)
  23          8 =   0x8 - enable debugging dumps of processes etc.
  24         16 =  0x10 - enable sync command
  25         32 =  0x20 - enable remount read-only
  26         64 =  0x40 - enable signalling of processes (term, kill, oom-kill)
  27        128 =  0x80 - allow reboot/poweroff
  28        256 = 0x100 - allow nicing of all RT tasks
  29
  30You can set the value in the file by the following command:
  31    echo "number" >/proc/sys/kernel/sysrq
  32
  33The number may be written here either as decimal or as hexadecimal
  34with the 0x prefix. CONFIG_MAGIC_SYSRQ_DEFAULT_ENABLE must always be
  35written in hexadecimal.
  36
  37Note that the value of /proc/sys/kernel/sysrq influences only the invocation
  38via a keyboard. Invocation of any operation via /proc/sysrq-trigger is always
  39allowed (by a user with admin privileges).
  40
  41*  How do I use the magic SysRq key?
  42~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
  43On x86   - You press the key combo 'ALT-SysRq-<command key>'. Note - Some
  44           keyboards may not have a key labeled 'SysRq'. The 'SysRq' key is
  45           also known as the 'Print Screen' key. Also some keyboards cannot
  46           handle so many keys being pressed at the same time, so you might
  47           have better luck with "press Alt", "press SysRq", "release SysRq",
  48           "press <command key>", release everything.
  49
  50On SPARC - You press 'ALT-STOP-<command key>', I believe.
  51
  52On the serial console (PC style standard serial ports only) -
  53           You send a BREAK, then within 5 seconds a command key. Sending
  54           BREAK twice is interpreted as a normal BREAK.
  55
  56On PowerPC - Press 'ALT - Print Screen (or F13) - <command key>,  
  57             Print Screen (or F13) - <command key> may suffice.
  58
  59On other - If you know of the key combos for other architectures, please
  60           let me know so I can add them to this section.
  61
  62On all -  write a character to /proc/sysrq-trigger.  e.g.:
  63
  64                echo t > /proc/sysrq-trigger
  65
  66*  What are the 'command' keys?
  67~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
  68'b'     - Will immediately reboot the system without syncing or unmounting
  69          your disks.
  70
  71'c'     - Will perform a system crash by a NULL pointer dereference.
  72          A crashdump will be taken if configured.
  73
  74'd'     - Shows all locks that are held.
  75
  76'e'     - Send a SIGTERM to all processes, except for init.
  77
  78'f'     - Will call oom_kill to kill a memory hog process.
  79
  80'g'     - Used by kgdb (kernel debugger)
  81
  82'h'     - Will display help (actually any other key than those listed
  83          here will display help. but 'h' is easy to remember :-)
  84
  85'i'     - Send a SIGKILL to all processes, except for init.
  86
  87'j'     - Forcibly "Just thaw it" - filesystems frozen by the FIFREEZE ioctl.
  88
  89'k'     - Secure Access Key (SAK) Kills all programs on the current virtual
  90          console. NOTE: See important comments below in SAK section.
  91
  92'l'     - Shows a stack backtrace for all active CPUs.
  93
  94'm'     - Will dump current memory info to your console.
  95
  96'n'     - Used to make RT tasks nice-able
  97
  98'o'     - Will shut your system off (if configured and supported).
  99
 100'p'     - Will dump the current registers and flags to your console.
 101
 102'q'     - Will dump per CPU lists of all armed hrtimers (but NOT regular
 103          timer_list timers) and detailed information about all
 104          clockevent devices.
 105
 106'r'     - Turns off keyboard raw mode and sets it to XLATE.
 107
 108's'     - Will attempt to sync all mounted filesystems.
 109
 110't'     - Will dump a list of current tasks and their information to your
 111          console.
 112
 113'u'     - Will attempt to remount all mounted filesystems read-only.
 114
 115'v'     - Forcefully restores framebuffer console
 116'v'     - Causes ETM buffer dump [ARM-specific]
 117
 118'w'     - Dumps tasks that are in uninterruptable (blocked) state.
 119
 120'x'     - Used by xmon interface on ppc/powerpc platforms.
 121          Show global PMU Registers on sparc64.
 122
 123'y'     - Show global CPU Registers [SPARC-64 specific]
 124
 125'z'     - Dump the ftrace buffer
 126
 127'0'-'9' - Sets the console log level, controlling which kernel messages
 128          will be printed to your console. ('0', for example would make
 129          it so that only emergency messages like PANICs or OOPSes would
 130          make it to your console.)
 131
 132*  Okay, so what can I use them for?
 133~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 134Well, unraw(r) is very handy when your X server or a svgalib program crashes.
 135
 136sak(k) (Secure Access Key) is useful when you want to be sure there is no
 137trojan program running at console which could grab your password
 138when you would try to login. It will kill all programs on given console,
 139thus letting you make sure that the login prompt you see is actually
 140the one from init, not some trojan program.
 141IMPORTANT: In its true form it is not a true SAK like the one in a :IMPORTANT
 142IMPORTANT: c2 compliant system, and it should not be mistaken as   :IMPORTANT
 143IMPORTANT: such.                                                   :IMPORTANT
 144       It seems others find it useful as (System Attention Key) which is
 145useful when you want to exit a program that will not let you switch consoles.
 146(For example, X or a svgalib program.)
 147
 148reboot(b) is good when you're unable to shut down. But you should also
 149sync(s) and umount(u) first.
 150
 151crash(c) can be used to manually trigger a crashdump when the system is hung.
 152Note that this just triggers a crash if there is no dump mechanism available.
 153
 154sync(s) is great when your system is locked up, it allows you to sync your
 155disks and will certainly lessen the chance of data loss and fscking. Note
 156that the sync hasn't taken place until you see the "OK" and "Done" appear
 157on the screen. (If the kernel is really in strife, you may not ever get the
 158OK or Done message...)
 159
 160umount(u) is basically useful in the same ways as sync(s). I generally sync(s),
 161umount(u), then reboot(b) when my system locks. It's saved me many a fsck.
 162Again, the unmount (remount read-only) hasn't taken place until you see the
 163"OK" and "Done" message appear on the screen.
 164
 165The loglevels '0'-'9' are useful when your console is being flooded with
 166kernel messages you do not want to see. Selecting '0' will prevent all but
 167the most urgent kernel messages from reaching your console. (They will
 168still be logged if syslogd/klogd are alive, though.)
 169
 170term(e) and kill(i) are useful if you have some sort of runaway process you
 171are unable to kill any other way, especially if it's spawning other
 172processes.
 173
 174"just thaw it(j)" is useful if your system becomes unresponsive due to a frozen
 175(probably root) filesystem via the FIFREEZE ioctl.
 176
 177*  Sometimes SysRq seems to get 'stuck' after using it, what can I do?
 178~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 179That happens to me, also. I've found that tapping shift, alt, and control
 180on both sides of the keyboard, and hitting an invalid sysrq sequence again
 181will fix the problem. (i.e., something like alt-sysrq-z). Switching to another
 182virtual console (ALT+Fn) and then back again should also help.
 183
 184*  I hit SysRq, but nothing seems to happen, what's wrong?
 185~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 186There are some keyboards that produce a different keycode for SysRq than the
 187pre-defined value of 99 (see KEY_SYSRQ in include/linux/input.h), or which
 188don't have a SysRq key at all. In these cases, run 'showkey -s' to find an
 189appropriate scancode sequence, and use 'setkeycodes <sequence> 99' to map
 190this sequence to the usual SysRq code (e.g., 'setkeycodes e05b 99'). It's
 191probably best to put this command in a boot script. Oh, and by the way, you
 192exit 'showkey' by not typing anything for ten seconds.
 193
 194*  I want to add SysRQ key events to a module, how does it work?
 195~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 196In order to register a basic function with the table, you must first include
 197the header 'include/linux/sysrq.h', this will define everything else you need.
 198Next, you must create a sysrq_key_op struct, and populate it with A) the key
 199handler function you will use, B) a help_msg string, that will print when SysRQ
 200prints help, and C) an action_msg string, that will print right before your
 201handler is called. Your handler must conform to the prototype in 'sysrq.h'.
 202
 203After the sysrq_key_op is created, you can call the kernel function
 204register_sysrq_key(int key, struct sysrq_key_op *op_p); this will
 205register the operation pointed to by 'op_p' at table key 'key',
 206if that slot in the table is blank. At module unload time, you must call
 207the function unregister_sysrq_key(int key, struct sysrq_key_op *op_p), which
 208will remove the key op pointed to by 'op_p' from the key 'key', if and only if
 209it is currently registered in that slot. This is in case the slot has been
 210overwritten since you registered it.
 211
 212The Magic SysRQ system works by registering key operations against a key op
 213lookup table, which is defined in 'drivers/char/sysrq.c'. This key table has
 214a number of operations registered into it at compile time, but is mutable,
 215and 2 functions are exported for interface to it:
 216        register_sysrq_key and unregister_sysrq_key.
 217Of course, never ever leave an invalid pointer in the table. I.e., when
 218your module that called register_sysrq_key() exits, it must call
 219unregister_sysrq_key() to clean up the sysrq key table entry that it used.
 220Null pointers in the table are always safe. :)
 221
 222If for some reason you feel the need to call the handle_sysrq function from
 223within a function called by handle_sysrq, you must be aware that you are in
 224a lock (you are also in an interrupt handler, which means don't sleep!), so
 225you must call __handle_sysrq_nolock instead.
 226
 227*  When I hit a SysRq key combination only the header appears on the console?
 228~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 229Sysrq output is subject to the same console loglevel control as all
 230other console output.  This means that if the kernel was booted 'quiet'
 231as is common on distro kernels the output may not appear on the actual
 232console, even though it will appear in the dmesg buffer, and be accessible
 233via the dmesg command and to the consumers of /proc/kmsg.  As a specific
 234exception the header line from the sysrq command is passed to all console
 235consumers as if the current loglevel was maximum.  If only the header
 236is emitted it is almost certain that the kernel loglevel is too low.
 237Should you require the output on the console channel then you will need
 238to temporarily up the console loglevel using alt-sysrq-8 or:
 239
 240    echo 8 > /proc/sysrq-trigger
 241
 242Remember to return the loglevel to normal after triggering the sysrq
 243command you are interested in.
 244
 245*  I have more questions, who can I ask?
 246~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 247Just ask them on the linux-kernel mailing list:
 248        linux-kernel@vger.kernel.org
 249
 250*  Credits
 251~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 252Written by Mydraal <vulpyne@vulpyne.net>
 253Updated by Adam Sulmicki <adam@cfar.umd.edu>
 254Updated by Jeremy M. Dolan <jmd@turbogeek.org> 2001/01/28 10:15:59
 255Added to by Crutcher Dunnavant <crutcher+kernel@datastacks.com>
 256
lxr.linux.no kindly hosted by Redpill Linpro AS, provider of Linux consulting and operations services since 1995.