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