1Linux Magic System Request Key Hacks
2====================================
3
4Documentation for sysrq.c
5
6What is the magic SysRq key?
7~~~~~~~~~~~~~~~~~~~~~~~~~~~~
8
9It is a 'magical' key combo you can hit which the kernel will respond to
10regardless of whatever else it is doing, unless it is completely locked up.
11
12How do I enable the magic SysRq key?
13~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
14
15You need to say "yes" to 'Magic SysRq key (CONFIG_MAGIC_SYSRQ)' when
16configuring the kernel. When running a kernel with SysRq compiled in,
17/proc/sys/kernel/sysrq controls the functions allowed to be invoked via
18the SysRq key. The default value in this file is set by the
19CONFIG_MAGIC_SYSRQ_DEFAULT_ENABLE config symbol, which itself defaults
20to 1. Here is the list of possible values in /proc/sys/kernel/sysrq:
21
22   -  0 - disable sysrq completely
23   -  1 - enable all functions of sysrq
24   - >1 - bitmask of allowed sysrq functions (see below for detailed function
25     description)::
26
27          2 =   0x2 - enable control of console logging level
28          4 =   0x4 - enable control of keyboard (SAK, unraw)
29          8 =   0x8 - enable debugging dumps of processes etc.
30         16 =  0x10 - enable sync command
31         32 =  0x20 - enable remount read-only
32         64 =  0x40 - enable signalling of processes (term, kill, oom-kill)
33        128 =  0x80 - allow reboot/poweroff
34        256 = 0x100 - allow nicing of all RT tasks
35
36You can set the value in the file by the following command::
37
38    echo "number" >/proc/sys/kernel/sysrq
39
40The number may be written here either as decimal or as hexadecimal
41with the 0x prefix. CONFIG_MAGIC_SYSRQ_DEFAULT_ENABLE must always be
42written in hexadecimal.
43
44Note that the value of ``/proc/sys/kernel/sysrq`` influences only the invocation
45via a keyboard. Invocation of any operation via ``/proc/sysrq-trigger`` is
46always allowed (by a user with admin privileges).
47
48How do I use the magic SysRq key?
49~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
50
51On x86   - You press the key combo :kbd:`ALT-SysRq-<command key>`.
52
53.. note::
54	   Some
55           keyboards may not have a key labeled 'SysRq'. The 'SysRq' key is
56           also known as the 'Print Screen' key. Also some keyboards cannot
57	   handle so many keys being pressed at the same time, so you might
58	   have better luck with press :kbd:`Alt`, press :kbd:`SysRq`,
59	   release :kbd:`SysRq`, press :kbd:`<command key>`, release everything.
60
61On SPARC - You press :kbd:`ALT-STOP-<command key>`, I believe.
62
63On the serial console (PC style standard serial ports only)
64        You send a ``BREAK``, then within 5 seconds a command key. Sending
65        ``BREAK`` twice is interpreted as a normal BREAK.
66
67On PowerPC
68	Press :kbd:`ALT - Print Screen` (or :kbd:`F13`) - :kbd:`<command key>`,
69        :kbd:`Print Screen` (or :kbd:`F13`) - :kbd:`<command key>` may suffice.
70
71On other
72	If you know of the key combos for other architectures, please
73        let me know so I can add them to this section.
74
75On all
76	write a character to /proc/sysrq-trigger.  e.g.::
77
78		echo t > /proc/sysrq-trigger
79
80What are the 'command' keys?
81~~~~~~~~~~~~~~~~~~~~~~~~~~~~
82
83=========== ===================================================================
84Command	    Function
85=========== ===================================================================
86``b``	    Will immediately reboot the system without syncing or unmounting
87            your disks.
88
89``c``	    Will perform a system crash by a NULL pointer dereference.
90            A crashdump will be taken if configured.
91
92``d``	    Shows all locks that are held.
93
94``e``	    Send a SIGTERM to all processes, except for init.
95
96``f``	    Will call the oom killer to kill a memory hog process, but do not
97	    panic if nothing can be killed.
98
99``g``	    Used by kgdb (kernel debugger)
100
101``h``	    Will display help (actually any other key than those listed
102            here will display help. but ``h`` is easy to remember :-)
103
104``i``	    Send a SIGKILL to all processes, except for init.
105
106``j``	    Forcibly "Just thaw it" - filesystems frozen by the FIFREEZE ioctl.
107
108``k``	    Secure Access Key (SAK) Kills all programs on the current virtual
109            console. NOTE: See important comments below in SAK section.
110
111``l``	    Shows a stack backtrace for all active CPUs.
112
113``m``	    Will dump current memory info to your console.
114
115``n``	    Used to make RT tasks nice-able
116
117``o``	    Will shut your system off (if configured and supported).
118
119``p``	    Will dump the current registers and flags to your console.
120
121``q``	    Will dump per CPU lists of all armed hrtimers (but NOT regular
122            timer_list timers) and detailed information about all
123            clockevent devices.
124
125``r``	    Turns off keyboard raw mode and sets it to XLATE.
126
127``s``	    Will attempt to sync all mounted filesystems.
128
129``t``	    Will dump a list of current tasks and their information to your
130            console.
131
132``u``	    Will attempt to remount all mounted filesystems read-only.
133
134``v``	    Forcefully restores framebuffer console
135``v``	    Causes ETM buffer dump [ARM-specific]
136
137``w``	    Dumps tasks that are in uninterruptable (blocked) state.
138
139``x``	    Used by xmon interface on ppc/powerpc platforms.
140            Show global PMU Registers on sparc64.
141            Dump all TLB entries on MIPS.
142
143``y``	    Show global CPU Registers [SPARC-64 specific]
144
145``z``	    Dump the ftrace buffer
146
147``0``-``9`` Sets the console log level, controlling which kernel messages
148            will be printed to your console. (``0``, for example would make
149            it so that only emergency messages like PANICs or OOPSes would
150            make it to your console.)
151=========== ===================================================================
152
153Okay, so what can I use them for?
154~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
155
156Well, unraw(r) is very handy when your X server or a svgalib program crashes.
157
158sak(k) (Secure Access Key) is useful when you want to be sure there is no
159trojan program running at console which could grab your password
160when you would try to login. It will kill all programs on given console,
161thus letting you make sure that the login prompt you see is actually
162the one from init, not some trojan program.
163
164.. important::
165
166   In its true form it is not a true SAK like the one in a
167   c2 compliant system, and it should not be mistaken as
168   such.
169
170It seems others find it useful as (System Attention Key) which is
171useful when you want to exit a program that will not let you switch consoles.
172(For example, X or a svgalib program.)
173
174``reboot(b)`` is good when you're unable to shut down, it is an equivalent
175of pressing the "reset" button.
176
177``crash(c)`` can be used to manually trigger a crashdump when the system is hung.
178Note that this just triggers a crash if there is no dump mechanism available.
179
180``sync(s)`` is handy before yanking removable medium or after using a rescue
181shell that provides no graceful shutdown -- it will ensure your data is
182safely written to the disk. Note that the sync hasn't taken place until you see
183the "OK" and "Done" appear on the screen.
184
185``umount(u)`` can be used to mark filesystems as properly unmounted. From the
186running system's point of view, they will be remounted read-only. The remount
187isn't complete until you see the "OK" and "Done" message appear on the screen.
188
189The loglevels ``0``-``9`` are useful when your console is being flooded with
190kernel messages you do not want to see. Selecting ``0`` will prevent all but
191the most urgent kernel messages from reaching your console. (They will
192still be logged if syslogd/klogd are alive, though.)
193
194``term(e)`` and ``kill(i)`` are useful if you have some sort of runaway process
195you are unable to kill any other way, especially if it's spawning other
196processes.
197
198"just thaw ``it(j)``" is useful if your system becomes unresponsive due to a
199frozen (probably root) filesystem via the FIFREEZE ioctl.
200
201Sometimes SysRq seems to get 'stuck' after using it, what can I do?
202~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
203
204That happens to me, also. I've found that tapping shift, alt, and control
205on both sides of the keyboard, and hitting an invalid sysrq sequence again
206will fix the problem. (i.e., something like :kbd:`alt-sysrq-z`). Switching to
207another virtual console (:kbd:`ALT+Fn`) and then back again should also help.
208
209I hit SysRq, but nothing seems to happen, what's wrong?
210~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
211
212There are some keyboards that produce a different keycode for SysRq than the
213pre-defined value of 99
214(see ``KEY_SYSRQ`` in ``include/uapi/linux/input-event-codes.h``), or
215which don't have a SysRq key at all. In these cases, run ``showkey -s`` to find
216an appropriate scancode sequence, and use ``setkeycodes <sequence> 99`` to map
217this sequence to the usual SysRq code (e.g., ``setkeycodes e05b 99``). It's
218probably best to put this command in a boot script. Oh, and by the way, you
219exit ``showkey`` by not typing anything for ten seconds.
220
221I want to add SysRQ key events to a module, how does it work?
222~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
223
224In order to register a basic function with the table, you must first include
225the header ``include/linux/sysrq.h``, this will define everything else you need.
226Next, you must create a ``sysrq_key_op`` struct, and populate it with A) the key
227handler function you will use, B) a help_msg string, that will print when SysRQ
228prints help, and C) an action_msg string, that will print right before your
229handler is called. Your handler must conform to the prototype in 'sysrq.h'.
230
231After the ``sysrq_key_op`` is created, you can call the kernel function
232``register_sysrq_key(int key, struct sysrq_key_op *op_p);`` this will
233register the operation pointed to by ``op_p`` at table key 'key',
234if that slot in the table is blank. At module unload time, you must call
235the function ``unregister_sysrq_key(int key, struct sysrq_key_op *op_p)``, which
236will remove the key op pointed to by 'op_p' from the key 'key', if and only if
237it is currently registered in that slot. This is in case the slot has been
238overwritten since you registered it.
239
240The Magic SysRQ system works by registering key operations against a key op
241lookup table, which is defined in 'drivers/tty/sysrq.c'. This key table has
242a number of operations registered into it at compile time, but is mutable,
243and 2 functions are exported for interface to it::
244
245	register_sysrq_key and unregister_sysrq_key.
246
247Of course, never ever leave an invalid pointer in the table. I.e., when
248your module that called register_sysrq_key() exits, it must call
249unregister_sysrq_key() to clean up the sysrq key table entry that it used.
250Null pointers in the table are always safe. :)
251
252If for some reason you feel the need to call the handle_sysrq function from
253within a function called by handle_sysrq, you must be aware that you are in
254a lock (you are also in an interrupt handler, which means don't sleep!), so
255you must call ``__handle_sysrq_nolock`` instead.
256
257When I hit a SysRq key combination only the header appears on the console?
258~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
259
260Sysrq output is subject to the same console loglevel control as all
261other console output.  This means that if the kernel was booted 'quiet'
262as is common on distro kernels the output may not appear on the actual
263console, even though it will appear in the dmesg buffer, and be accessible
264via the dmesg command and to the consumers of ``/proc/kmsg``.  As a specific
265exception the header line from the sysrq command is passed to all console
266consumers as if the current loglevel was maximum.  If only the header
267is emitted it is almost certain that the kernel loglevel is too low.
268Should you require the output on the console channel then you will need
269to temporarily up the console loglevel using :kbd:`alt-sysrq-8` or::
270
271    echo 8 > /proc/sysrq-trigger
272
273Remember to return the loglevel to normal after triggering the sysrq
274command you are interested in.
275
276I have more questions, who can I ask?
277~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
278
279Just ask them on the linux-kernel mailing list:
280	linux-kernel@vger.kernel.org
281
282Credits
283~~~~~~~
284
285Written by Mydraal <vulpyne@vulpyne.net>
286Updated by Adam Sulmicki <adam@cfar.umd.edu>
287Updated by Jeremy M. Dolan <jmd@turbogeek.org> 2001/01/28 10:15:59
288Added to by Crutcher Dunnavant <crutcher+kernel@datastacks.com>
289