ia64/linux-2.6.18-xen.hg

view Documentation/sysrq.txt @ 897:329ea0ccb344

balloon: try harder to balloon up under memory pressure.

Currently if the balloon driver is unable to increase the guest's
reservation it assumes the failure was due to reaching its full
allocation, gives up on the ballooning operation and records the limit
it reached as the "hard limit". The driver will not try again until
the target is set again (even to the same value).

However it is possible that ballooning has in fact failed due to
memory pressure in the host and therefore it is desirable to keep
attempting to reach the target in case memory becomes available. The
most likely scenario is that some guests are ballooning down while
others are ballooning up and therefore there is temporary memory
pressure while things stabilise. You would not expect a well behaved
toolstack to ask a domain to balloon to more than its allocation nor
would you expect it to deliberately over-commit memory by setting
balloon targets which exceed the total host memory.

This patch drops the concept of a hard limit and causes the balloon
driver to retry increasing the reservation on a timer in the same
manner as when decreasing the reservation.

Also if we partially succeed in increasing the reservation
(i.e. receive less pages than we asked for) then we may as well keep
those pages rather than returning them to Xen.

Signed-off-by: Ian Campbell <ian.campbell@citrix.com>
author Keir Fraser <keir.fraser@citrix.com>
date Fri Jun 05 14:01:20 2009 +0100 (2009-06-05)
parents 831230e53067
children
line source
1 Linux Magic System Request Key Hacks
2 Documentation for sysrq.c version 1.15
3 Last update: $Date: 2001/01/28 10:15:59 $
5 * What is the magic SysRq key?
6 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
7 It is a 'magical' key combo you can hit which the kernel will respond to
8 regardless of whatever else it is doing, unless it is completely locked up.
10 * How do I enable the magic SysRq key?
11 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
12 You need to say "yes" to 'Magic SysRq key (CONFIG_MAGIC_SYSRQ)' when
13 configuring the kernel. When running a kernel with SysRq compiled in,
14 /proc/sys/kernel/sysrq controls the functions allowed to be invoked via
15 the SysRq key. By default the file contains 1 which means that every
16 possible SysRq request is allowed (in older versions SysRq was disabled
17 by default, and you were required to specifically enable it at run-time
18 but this is not the case any more). Here is the list of possible values
19 in /proc/sys/kernel/sysrq:
20 0 - disable sysrq completely
21 1 - enable all functions of sysrq
22 >1 - bitmask of allowed sysrq functions (see below for detailed function
23 description):
24 2 - enable control of console logging level
25 4 - enable control of keyboard (SAK, unraw)
26 8 - enable debugging dumps of processes etc.
27 16 - enable sync command
28 32 - enable remount read-only
29 64 - enable signalling of processes (term, kill, oom-kill)
30 128 - allow reboot/poweroff
31 256 - allow nicing of all RT tasks
33 You can set the value in the file by the following command:
34 echo "number" >/proc/sys/kernel/sysrq
36 Note that the value of /proc/sys/kernel/sysrq influences only the invocation
37 via a keyboard. Invocation of any operation via /proc/sysrq-trigger is always
38 allowed.
40 * How do I use the magic SysRq key?
41 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
42 On x86 - You press the key combo 'ALT-SysRq-<command key>'. Note - Some
43 keyboards may not have a key labeled 'SysRq'. The 'SysRq' key is
44 also known as the 'Print Screen' key. Also some keyboards cannot
45 handle so many keys being pressed at the same time, so you might
46 have better luck with "press Alt", "press SysRq", "release Alt",
47 "press <command key>", release everything.
49 On SPARC - You press 'ALT-STOP-<command key>', I believe.
51 On the serial console (PC style standard serial ports only) -
52 You send a BREAK, then within 5 seconds a command key. Sending
53 BREAK twice is interpreted as a normal BREAK.
55 On PowerPC - Press 'ALT - Print Screen (or F13) - <command key>,
56 Print Screen (or F13) - <command key> may suffice.
58 On other - If you know of the key combos for other architectures, please
59 let me know so I can add them to this section.
61 On all - write a character to /proc/sysrq-trigger. eg:
63 echo t > /proc/sysrq-trigger
65 * What are the 'command' keys?
66 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
67 'r' - Turns off keyboard raw mode and sets it to XLATE.
69 'k' - Secure Access Key (SAK) Kills all programs on the current virtual
70 console. NOTE: See important comments below in SAK section.
72 'b' - Will immediately reboot the system without syncing or unmounting
73 your disks.
75 'c' - Will perform a kexec reboot in order to take a crashdump.
77 'o' - Will shut your system off (if configured and supported).
79 's' - Will attempt to sync all mounted filesystems.
81 'u' - Will attempt to remount all mounted filesystems read-only.
83 'p' - Will dump the current registers and flags to your console.
85 't' - Will dump a list of current tasks and their information to your
86 console.
88 'm' - Will dump current memory info to your console.
90 'v' - Dumps Voyager SMP processor info to your console.
92 '0'-'9' - Sets the console log level, controlling which kernel messages
93 will be printed to your console. ('0', for example would make
94 it so that only emergency messages like PANICs or OOPSes would
95 make it to your console.)
97 'f' - Will call oom_kill to kill a memory hog process
99 'e' - Send a SIGTERM to all processes, except for init.
101 'i' - Send a SIGKILL to all processes, except for init.
103 'l' - Send a SIGKILL to all processes, INCLUDING init. (Your system
104 will be non-functional after this.)
106 'h' - Will display help ( actually any other key than those listed
107 above will display help. but 'h' is easy to remember :-)
109 * Okay, so what can I use them for?
110 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
111 Well, un'R'aw is very handy when your X server or a svgalib program crashes.
113 sa'K' (Secure Access Key) is useful when you want to be sure there are no
114 trojan program is running at console and which could grab your password
115 when you would try to login. It will kill all programs on given console
116 and thus letting you make sure that the login prompt you see is actually
117 the one from init, not some trojan program.
118 IMPORTANT: In its true form it is not a true SAK like the one in a :IMPORTANT
119 IMPORTANT: c2 compliant system, and it should not be mistaken as :IMPORTANT
120 IMPORTANT: such. :IMPORTANT
121 It seems other find it useful as (System Attention Key) which is
122 useful when you want to exit a program that will not let you switch consoles.
123 (For example, X or a svgalib program.)
125 re'B'oot is good when you're unable to shut down. But you should also 'S'ync
126 and 'U'mount first.
128 'C'rashdump can be used to manually trigger a crashdump when the system is hung.
129 The kernel needs to have been built with CONFIG_KEXEC enabled.
131 'S'ync is great when your system is locked up, it allows you to sync your
132 disks and will certainly lessen the chance of data loss and fscking. Note
133 that the sync hasn't taken place until you see the "OK" and "Done" appear
134 on the screen. (If the kernel is really in strife, you may not ever get the
135 OK or Done message...)
137 'U'mount is basically useful in the same ways as 'S'ync. I generally 'S'ync,
138 'U'mount, then re'B'oot when my system locks. It's saved me many a fsck.
139 Again, the unmount (remount read-only) hasn't taken place until you see the
140 "OK" and "Done" message appear on the screen.
142 The loglevel'0'-'9' is useful when your console is being flooded with
143 kernel messages you do not want to see. Setting '0' will prevent all but
144 the most urgent kernel messages from reaching your console. (They will
145 still be logged if syslogd/klogd are alive, though.)
147 t'E'rm and k'I'll are useful if you have some sort of runaway process you
148 are unable to kill any other way, especially if it's spawning other
149 processes.
151 * Sometimes SysRq seems to get 'stuck' after using it, what can I do?
152 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
153 That happens to me, also. I've found that tapping shift, alt, and control
154 on both sides of the keyboard, and hitting an invalid sysrq sequence again
155 will fix the problem. (ie, something like alt-sysrq-z). Switching to another
156 virtual console (ALT+Fn) and then back again should also help.
158 * I hit SysRq, but nothing seems to happen, what's wrong?
159 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
160 There are some keyboards that send different scancodes for SysRq than the
161 pre-defined 0x54. So if SysRq doesn't work out of the box for a certain
162 keyboard, run 'showkey -s' to find out the proper scancode sequence. Then
163 use 'setkeycodes <sequence> 84' to define this sequence to the usual SysRq
164 code (84 is decimal for 0x54). It's probably best to put this command in a
165 boot script. Oh, and by the way, you exit 'showkey' by not typing anything
166 for ten seconds.
168 * I want to add SysRQ key events to a module, how does it work?
169 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
170 In order to register a basic function with the table, you must first include
171 the header 'include/linux/sysrq.h', this will define everything else you need.
172 Next, you must create a sysrq_key_op struct, and populate it with A) the key
173 handler function you will use, B) a help_msg string, that will print when SysRQ
174 prints help, and C) an action_msg string, that will print right before your
175 handler is called. Your handler must conform to the prototype in 'sysrq.h'.
177 After the sysrq_key_op is created, you can call the macro
178 register_sysrq_key(int key, struct sysrq_key_op *op_p) that is defined in
179 sysrq.h, this will register the operation pointed to by 'op_p' at table
180 key 'key', if that slot in the table is blank. At module unload time, you must
181 call the macro unregister_sysrq_key(int key, struct sysrq_key_op *op_p), which
182 will remove the key op pointed to by 'op_p' from the key 'key', if and only if
183 it is currently registered in that slot. This is in case the slot has been
184 overwritten since you registered it.
186 The Magic SysRQ system works by registering key operations against a key op
187 lookup table, which is defined in 'drivers/char/sysrq.c'. This key table has
188 a number of operations registered into it at compile time, but is mutable,
189 and 4 functions are exported for interface to it: __sysrq_lock_table,
190 __sysrq_unlock_table, __sysrq_get_key_op, and __sysrq_put_key_op. The
191 functions __sysrq_swap_key_ops and __sysrq_swap_key_ops_nolock are defined
192 in the header itself, and the REGISTER and UNREGISTER macros are built from
193 these. More complex (and dangerous!) manipulations of the table are possible
194 using these functions, but you must be careful to always lock the table before
195 you read or write from it, and to unlock it again when you are done. (And of
196 course, to never ever leave an invalid pointer in the table). Null pointers in
197 the table are always safe :)
199 If for some reason you feel the need to call the handle_sysrq function from
200 within a function called by handle_sysrq, you must be aware that you are in
201 a lock (you are also in an interrupt handler, which means don't sleep!), so
202 you must call __handle_sysrq_nolock instead.
204 * I have more questions, who can I ask?
205 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
206 And I'll answer any questions about the registration system you got, also
207 responding as soon as possible.
208 -Crutcher
210 * Credits
211 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
212 Written by Mydraal <vulpyne@vulpyne.net>
213 Updated by Adam Sulmicki <adam@cfar.umd.edu>
214 Updated by Jeremy M. Dolan <jmd@turbogeek.org> 2001/01/28 10:15:59
215 Added to by Crutcher Dunnavant <crutcher+kernel@datastacks.com>