blob: 22dd6af2e4bd42152edbe872b224b85a769e7184 [file] [log] [blame]
Fernando Luis Vázquez Cao9919cba2012-02-09 17:42:20 -05001===============================================================
2Softlockup detector and hardlockup detector (aka nmi_watchdog)
3===============================================================
4
5The Linux kernel can act as a watchdog to detect both soft and hard
6lockups.
7
8A 'softlockup' is defined as a bug that causes the kernel to loop in
9kernel mode for more than 20 seconds (see "Implementation" below for
10details), without giving other tasks a chance to run. The current
11stack trace is displayed upon detection and, by default, the system
12will stay locked up. Alternatively, the kernel can be configured to
13panic; a sysctl, "kernel.softlockup_panic", a kernel parameter,
14"softlockup_panic" (see "Documentation/kernel-parameters.txt" for
Harish Jenny K N8ae34ea2014-08-20 11:56:17 +053015details), and a compile option, "BOOTPARAM_SOFTLOCKUP_PANIC", are
Fernando Luis Vázquez Cao9919cba2012-02-09 17:42:20 -050016provided for this.
17
18A 'hardlockup' is defined as a bug that causes the CPU to loop in
19kernel mode for more than 10 seconds (see "Implementation" below for
20details), without letting other interrupts have a chance to run.
21Similarly to the softlockup case, the current stack trace is displayed
22upon detection and the system will stay locked up unless the default
23behavior is changed, which can be done through a compile time knob,
24"BOOTPARAM_HARDLOCKUP_PANIC", and a kernel parameter, "nmi_watchdog"
25(see "Documentation/kernel-parameters.txt" for details).
26
27The panic option can be used in combination with panic_timeout (this
28timeout is set through the confusingly named "kernel.panic" sysctl),
29to cause the system to reboot automatically after a specified amount
30of time.
31
32=== Implementation ===
33
34The soft and hard lockup detectors are built on top of the hrtimer and
35perf subsystems, respectively. A direct consequence of this is that,
36in principle, they should work in any architecture where these
37subsystems are present.
38
39A periodic hrtimer runs to generate interrupts and kick the watchdog
40task. An NMI perf event is generated every "watchdog_thresh"
41(compile-time initialized to 10 and configurable through sysctl of the
42same name) seconds to check for hardlockups. If any CPU in the system
43does not receive any hrtimer interrupt during that time the
44'hardlockup detector' (the handler for the NMI perf event) will
45generate a kernel warning or call panic, depending on the
46configuration.
47
48The watchdog task is a high priority kernel thread that updates a
49timestamp every time it is scheduled. If that timestamp is not updated
50for 2*watchdog_thresh seconds (the softlockup threshold) the
51'softlockup detector' (coded inside the hrtimer callback function)
52will dump useful debug information to the system log, after which it
53will call panic if it was instructed to do so or resume execution of
54other kernel code.
55
56The period of the hrtimer is 2*watchdog_thresh/5, which means it has
57two or three chances to generate an interrupt before the hardlockup
58detector kicks in.
59
60As explained above, a kernel knob is provided that allows
61administrators to configure the period of the hrtimer and the perf
62event. The right value for a particular environment is a trade-off
63between fast response to lockups and detection overhead.
Chris Metcalffe4ba3c2015-06-24 16:55:45 -070064
65By default, the watchdog runs on all online cores. However, on a
66kernel configured with NO_HZ_FULL, by default the watchdog runs only
67on the housekeeping cores, not the cores specified in the "nohz_full"
68boot argument. If we allowed the watchdog to run by default on
69the "nohz_full" cores, we would have to run timer ticks to activate
70the scheduler, which would prevent the "nohz_full" functionality
71from protecting the user code on those cores from the kernel.
72Of course, disabling it by default on the nohz_full cores means that
73when those cores do enter the kernel, by default we will not be
74able to detect if they lock up. However, allowing the watchdog
75to continue to run on the housekeeping (non-tickless) cores means
76that we will continue to detect lockups properly on those cores.
77
78In either case, the set of cores excluded from running the watchdog
79may be adjusted via the kernel.watchdog_cpumask sysctl. For
80nohz_full cores, this may be useful for debugging a case where the
81kernel seems to be hanging on the nohz_full cores.