2010-05-08 01:11:44 +04:00
|
|
|
/*
|
|
|
|
* Detect hard and soft lockups on a system
|
|
|
|
*
|
|
|
|
* started by Don Zickus, Copyright (C) 2010 Red Hat, Inc.
|
|
|
|
*
|
2012-02-10 02:42:22 +04:00
|
|
|
* Note: Most of this code is borrowed heavily from the original softlockup
|
|
|
|
* detector, so thanks to Ingo for the initial implementation.
|
|
|
|
* Some chunks also taken from the old x86-specific nmi watchdog code, thanks
|
2010-05-08 01:11:44 +04:00
|
|
|
* to those contributors as well.
|
|
|
|
*/
|
|
|
|
|
2017-07-15 00:49:46 +03:00
|
|
|
#define pr_fmt(fmt) "watchdog: " fmt
|
2012-03-24 02:01:55 +04:00
|
|
|
|
2010-05-08 01:11:44 +04:00
|
|
|
#include <linux/mm.h>
|
|
|
|
#include <linux/cpu.h>
|
|
|
|
#include <linux/nmi.h>
|
|
|
|
#include <linux/init.h>
|
|
|
|
#include <linux/module.h>
|
|
|
|
#include <linux/sysctl.h>
|
2012-07-16 14:42:38 +04:00
|
|
|
#include <linux/smpboot.h>
|
2013-02-07 19:47:07 +04:00
|
|
|
#include <linux/sched/rt.h>
|
2017-02-01 20:07:51 +03:00
|
|
|
#include <uapi/linux/sched/types.h>
|
2015-06-25 02:55:45 +03:00
|
|
|
#include <linux/tick.h>
|
workqueue: implement lockup detector
Workqueue stalls can happen from a variety of usage bugs such as
missing WQ_MEM_RECLAIM flag or concurrency managed work item
indefinitely staying RUNNING. These stalls can be extremely difficult
to hunt down because the usual warning mechanisms can't detect
workqueue stalls and the internal state is pretty opaque.
To alleviate the situation, this patch implements workqueue lockup
detector. It periodically monitors all worker_pools periodically and,
if any pool failed to make forward progress longer than the threshold
duration, triggers warning and dumps workqueue state as follows.
BUG: workqueue lockup - pool cpus=0 node=0 flags=0x0 nice=0 stuck for 31s!
Showing busy workqueues and worker pools:
workqueue events: flags=0x0
pwq 0: cpus=0 node=0 flags=0x0 nice=0 active=17/256
pending: monkey_wrench_fn, e1000_watchdog, cache_reap, vmstat_shepherd, release_one_tty, release_one_tty, release_one_tty, release_one_tty, release_one_tty, release_one_tty, release_one_tty, release_one_tty, release_one_tty, release_one_tty, release_one_tty, release_one_tty, cgroup_release_agent
workqueue events_power_efficient: flags=0x80
pwq 0: cpus=0 node=0 flags=0x0 nice=0 active=2/256
pending: check_lifetime, neigh_periodic_work
workqueue cgroup_pidlist_destroy: flags=0x0
pwq 0: cpus=0 node=0 flags=0x0 nice=0 active=1/1
pending: cgroup_pidlist_destroy_work_fn
...
The detection mechanism is controller through kernel parameter
workqueue.watchdog_thresh and can be updated at runtime through the
sysfs module parameter file.
v2: Decoupled from softlockup control knobs.
Signed-off-by: Tejun Heo <tj@kernel.org>
Acked-by: Don Zickus <dzickus@redhat.com>
Cc: Ulrich Obergfell <uobergfe@redhat.com>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Chris Mason <clm@fb.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
2015-12-08 19:28:04 +03:00
|
|
|
#include <linux/workqueue.h>
|
2017-02-01 18:36:40 +03:00
|
|
|
#include <linux/sched/clock.h>
|
2017-02-08 20:51:35 +03:00
|
|
|
#include <linux/sched/debug.h>
|
2010-05-08 01:11:44 +04:00
|
|
|
|
|
|
|
#include <asm/irq_regs.h>
|
2012-03-10 23:37:28 +04:00
|
|
|
#include <linux/kvm_para.h>
|
2015-09-05 01:45:15 +03:00
|
|
|
#include <linux/kthread.h>
|
2010-05-08 01:11:44 +04:00
|
|
|
|
2017-09-12 22:37:01 +03:00
|
|
|
static DEFINE_MUTEX(watchdog_mutex);
|
2015-05-18 12:31:50 +03:00
|
|
|
|
2017-07-13 00:35:46 +03:00
|
|
|
#if defined(CONFIG_HARDLOCKUP_DETECTOR) || defined(CONFIG_HAVE_NMI_WATCHDOG)
|
2017-09-12 22:37:17 +03:00
|
|
|
# define WATCHDOG_DEFAULT (SOFT_WATCHDOG_ENABLED | NMI_WATCHDOG_ENABLED)
|
|
|
|
# define NMI_WATCHDOG_DEFAULT 1
|
2015-04-15 01:43:55 +03:00
|
|
|
#else
|
2017-09-12 22:37:17 +03:00
|
|
|
# define WATCHDOG_DEFAULT (SOFT_WATCHDOG_ENABLED)
|
|
|
|
# define NMI_WATCHDOG_DEFAULT 0
|
2015-04-15 01:43:55 +03:00
|
|
|
#endif
|
2017-07-13 00:35:46 +03:00
|
|
|
|
2017-09-12 22:37:17 +03:00
|
|
|
unsigned long __read_mostly watchdog_enabled;
|
|
|
|
int __read_mostly watchdog_user_enabled = 1;
|
|
|
|
int __read_mostly nmi_watchdog_user_enabled = NMI_WATCHDOG_DEFAULT;
|
|
|
|
int __read_mostly soft_watchdog_user_enabled = 1;
|
2017-09-12 22:37:15 +03:00
|
|
|
int __read_mostly watchdog_thresh = 10;
|
2017-09-12 22:37:19 +03:00
|
|
|
int __read_mostly nmi_watchdog_available;
|
2017-09-12 22:37:15 +03:00
|
|
|
|
|
|
|
struct cpumask watchdog_allowed_mask __read_mostly;
|
|
|
|
static bool softlockup_threads_initialized __read_mostly;
|
|
|
|
|
|
|
|
struct cpumask watchdog_cpumask __read_mostly;
|
|
|
|
unsigned long *watchdog_cpumask_bits = cpumask_bits(&watchdog_cpumask);
|
|
|
|
|
2017-07-13 00:35:46 +03:00
|
|
|
#ifdef CONFIG_HARDLOCKUP_DETECTOR
|
|
|
|
/*
|
|
|
|
* Should we panic when a soft-lockup or hard-lockup occurs:
|
|
|
|
*/
|
|
|
|
unsigned int __read_mostly hardlockup_panic =
|
|
|
|
CONFIG_BOOTPARAM_HARDLOCKUP_PANIC_VALUE;
|
|
|
|
/*
|
|
|
|
* We may not want to enable hard lockup detection by default in all cases,
|
|
|
|
* for example when running the kernel as a guest on a hypervisor. In these
|
|
|
|
* cases this function can be called to disable hard lockup detection. This
|
|
|
|
* function should only be executed once by the boot processor before the
|
|
|
|
* kernel command line parameters are parsed, because otherwise it is not
|
|
|
|
* possible to override this in hardlockup_panic_setup().
|
|
|
|
*/
|
2017-09-12 22:37:02 +03:00
|
|
|
void __init hardlockup_detector_disable(void)
|
2017-07-13 00:35:46 +03:00
|
|
|
{
|
2017-09-12 22:37:17 +03:00
|
|
|
nmi_watchdog_user_enabled = 0;
|
2017-07-13 00:35:46 +03:00
|
|
|
}
|
|
|
|
|
|
|
|
static int __init hardlockup_panic_setup(char *str)
|
|
|
|
{
|
|
|
|
if (!strncmp(str, "panic", 5))
|
|
|
|
hardlockup_panic = 1;
|
|
|
|
else if (!strncmp(str, "nopanic", 7))
|
|
|
|
hardlockup_panic = 0;
|
|
|
|
else if (!strncmp(str, "0", 1))
|
2017-09-12 22:37:17 +03:00
|
|
|
nmi_watchdog_user_enabled = 0;
|
2017-07-13 00:35:46 +03:00
|
|
|
else if (!strncmp(str, "1", 1))
|
2017-09-12 22:37:17 +03:00
|
|
|
nmi_watchdog_user_enabled = 1;
|
2017-07-13 00:35:46 +03:00
|
|
|
return 1;
|
|
|
|
}
|
|
|
|
__setup("nmi_watchdog=", hardlockup_panic_setup);
|
|
|
|
|
2017-09-12 22:37:07 +03:00
|
|
|
# ifdef CONFIG_SMP
|
|
|
|
int __read_mostly sysctl_hardlockup_all_cpu_backtrace;
|
2017-07-13 00:35:46 +03:00
|
|
|
|
2017-09-12 22:37:07 +03:00
|
|
|
static int __init hardlockup_all_cpu_backtrace_setup(char *str)
|
|
|
|
{
|
|
|
|
sysctl_hardlockup_all_cpu_backtrace = !!simple_strtol(str, NULL, 0);
|
|
|
|
return 1;
|
|
|
|
}
|
|
|
|
__setup("hardlockup_all_cpu_backtrace=", hardlockup_all_cpu_backtrace_setup);
|
|
|
|
# endif /* CONFIG_SMP */
|
|
|
|
#endif /* CONFIG_HARDLOCKUP_DETECTOR */
|
2017-07-13 00:35:46 +03:00
|
|
|
|
|
|
|
/*
|
|
|
|
* These functions can be overridden if an architecture implements its
|
|
|
|
* own hardlockup detector.
|
2017-07-13 00:35:49 +03:00
|
|
|
*
|
|
|
|
* watchdog_nmi_enable/disable can be implemented to start and stop when
|
|
|
|
* softlockup watchdog threads start and stop. The arch must select the
|
|
|
|
* SOFTLOCKUP_DETECTOR Kconfig.
|
2017-07-13 00:35:46 +03:00
|
|
|
*/
|
|
|
|
int __weak watchdog_nmi_enable(unsigned int cpu)
|
|
|
|
{
|
|
|
|
return 0;
|
|
|
|
}
|
2017-09-12 22:37:04 +03:00
|
|
|
|
2017-07-13 00:35:46 +03:00
|
|
|
void __weak watchdog_nmi_disable(unsigned int cpu)
|
|
|
|
{
|
2017-09-12 22:37:04 +03:00
|
|
|
hardlockup_detector_perf_disable();
|
2017-07-13 00:35:46 +03:00
|
|
|
}
|
|
|
|
|
2017-09-12 22:37:19 +03:00
|
|
|
/* Return 0, if a NMI watchdog is available. Error code otherwise */
|
|
|
|
int __weak __init watchdog_nmi_probe(void)
|
|
|
|
{
|
|
|
|
return hardlockup_detector_perf_init();
|
|
|
|
}
|
|
|
|
|
2017-09-12 22:37:16 +03:00
|
|
|
/**
|
|
|
|
* watchdog_nmi_reconfigure - Optional function to reconfigure NMI watchdogs
|
|
|
|
* @run: If false stop the watchdogs on all enabled CPUs
|
|
|
|
* If true start the watchdogs on all enabled CPUs
|
|
|
|
*
|
|
|
|
* The core call order is:
|
|
|
|
* watchdog_nmi_reconfigure(false);
|
|
|
|
* update_variables();
|
|
|
|
* watchdog_nmi_reconfigure(true);
|
|
|
|
*
|
|
|
|
* The second call which starts the watchdogs again guarantees that the
|
|
|
|
* following variables are stable across the call.
|
2017-09-12 22:37:15 +03:00
|
|
|
* - watchdog_enabled
|
2017-07-13 00:35:49 +03:00
|
|
|
* - watchdog_thresh
|
|
|
|
* - watchdog_cpumask
|
2017-09-12 22:37:16 +03:00
|
|
|
*
|
|
|
|
* After the call the variables can be changed again.
|
2017-07-13 00:35:49 +03:00
|
|
|
*/
|
2017-09-12 22:37:16 +03:00
|
|
|
void __weak watchdog_nmi_reconfigure(bool run) { }
|
2017-07-13 00:35:49 +03:00
|
|
|
|
2017-09-12 22:37:17 +03:00
|
|
|
/**
|
|
|
|
* lockup_detector_update_enable - Update the sysctl enable bit
|
|
|
|
*
|
|
|
|
* Caller needs to make sure that the NMI/perf watchdogs are off, so this
|
|
|
|
* can't race with watchdog_nmi_disable().
|
|
|
|
*/
|
|
|
|
static void lockup_detector_update_enable(void)
|
|
|
|
{
|
|
|
|
watchdog_enabled = 0;
|
|
|
|
if (!watchdog_user_enabled)
|
|
|
|
return;
|
2017-09-12 22:37:19 +03:00
|
|
|
if (nmi_watchdog_available && nmi_watchdog_user_enabled)
|
2017-09-12 22:37:17 +03:00
|
|
|
watchdog_enabled |= NMI_WATCHDOG_ENABLED;
|
|
|
|
if (soft_watchdog_user_enabled)
|
|
|
|
watchdog_enabled |= SOFT_WATCHDOG_ENABLED;
|
|
|
|
}
|
|
|
|
|
2017-07-13 00:35:46 +03:00
|
|
|
#ifdef CONFIG_SOFTLOCKUP_DETECTOR
|
|
|
|
|
2017-09-12 22:37:06 +03:00
|
|
|
/* Global variables, exported for sysctl */
|
|
|
|
unsigned int __read_mostly softlockup_panic =
|
|
|
|
CONFIG_BOOTPARAM_SOFTLOCKUP_PANIC_VALUE;
|
2017-09-12 22:37:10 +03:00
|
|
|
|
2012-12-18 03:59:50 +04:00
|
|
|
static u64 __read_mostly sample_period;
|
2010-05-08 01:11:44 +04:00
|
|
|
|
|
|
|
static DEFINE_PER_CPU(unsigned long, watchdog_touch_ts);
|
|
|
|
static DEFINE_PER_CPU(struct task_struct *, softlockup_watchdog);
|
|
|
|
static DEFINE_PER_CPU(struct hrtimer, watchdog_hrtimer);
|
|
|
|
static DEFINE_PER_CPU(bool, softlockup_touch_sync);
|
|
|
|
static DEFINE_PER_CPU(bool, soft_watchdog_warn);
|
2012-07-16 14:42:38 +04:00
|
|
|
static DEFINE_PER_CPU(unsigned long, hrtimer_interrupts);
|
|
|
|
static DEFINE_PER_CPU(unsigned long, soft_lockup_hrtimer_cnt);
|
2014-10-10 02:25:17 +04:00
|
|
|
static DEFINE_PER_CPU(struct task_struct *, softlockup_task_ptr_saved);
|
2010-05-08 01:11:44 +04:00
|
|
|
static DEFINE_PER_CPU(unsigned long, hrtimer_interrupts_saved);
|
2014-06-24 00:22:05 +04:00
|
|
|
static unsigned long soft_lockup_nmi_warn;
|
2010-05-08 01:11:44 +04:00
|
|
|
|
|
|
|
static int __init softlockup_panic_setup(char *str)
|
|
|
|
{
|
|
|
|
softlockup_panic = simple_strtoul(str, NULL, 0);
|
|
|
|
return 1;
|
|
|
|
}
|
|
|
|
__setup("softlockup_panic=", softlockup_panic_setup);
|
|
|
|
|
|
|
|
static int __init nowatchdog_setup(char *str)
|
|
|
|
{
|
2017-09-12 22:37:17 +03:00
|
|
|
watchdog_user_enabled = 0;
|
2010-05-08 01:11:44 +04:00
|
|
|
return 1;
|
|
|
|
}
|
|
|
|
__setup("nowatchdog", nowatchdog_setup);
|
|
|
|
|
|
|
|
static int __init nosoftlockup_setup(char *str)
|
|
|
|
{
|
2017-09-12 22:37:17 +03:00
|
|
|
soft_watchdog_user_enabled = 0;
|
2010-05-08 01:11:44 +04:00
|
|
|
return 1;
|
|
|
|
}
|
|
|
|
__setup("nosoftlockup", nosoftlockup_setup);
|
2015-04-15 01:44:13 +03:00
|
|
|
|
2014-06-24 00:22:05 +04:00
|
|
|
#ifdef CONFIG_SMP
|
2017-09-12 22:37:07 +03:00
|
|
|
int __read_mostly sysctl_softlockup_all_cpu_backtrace;
|
|
|
|
|
2014-06-24 00:22:05 +04:00
|
|
|
static int __init softlockup_all_cpu_backtrace_setup(char *str)
|
|
|
|
{
|
2017-09-12 22:37:07 +03:00
|
|
|
sysctl_softlockup_all_cpu_backtrace = !!simple_strtol(str, NULL, 0);
|
2014-06-24 00:22:05 +04:00
|
|
|
return 1;
|
|
|
|
}
|
|
|
|
__setup("softlockup_all_cpu_backtrace=", softlockup_all_cpu_backtrace_setup);
|
2017-07-13 00:35:46 +03:00
|
|
|
#endif
|
2010-05-08 01:11:44 +04:00
|
|
|
|
2017-09-12 22:37:04 +03:00
|
|
|
static void __lockup_detector_cleanup(void);
|
|
|
|
|
2011-05-23 09:10:23 +04:00
|
|
|
/*
|
|
|
|
* Hard-lockup warnings should be triggered after just a few seconds. Soft-
|
|
|
|
* lockups can have false positives under extreme conditions. So we generally
|
|
|
|
* want a higher threshold for soft lockups than for hard lockups. So we couple
|
|
|
|
* the thresholds with a factor: we make the soft threshold twice the amount of
|
|
|
|
* time the hard threshold is.
|
|
|
|
*/
|
2011-05-24 07:43:18 +04:00
|
|
|
static int get_softlockup_thresh(void)
|
2011-05-23 09:10:23 +04:00
|
|
|
{
|
|
|
|
return watchdog_thresh * 2;
|
|
|
|
}
|
2010-05-08 01:11:44 +04:00
|
|
|
|
|
|
|
/*
|
|
|
|
* Returns seconds, approximately. We don't need nanosecond
|
|
|
|
* resolution, and we don't need to waste time with a big divide when
|
|
|
|
* 2^30ns == 1.074s.
|
|
|
|
*/
|
2012-12-27 06:49:44 +04:00
|
|
|
static unsigned long get_timestamp(void)
|
2010-05-08 01:11:44 +04:00
|
|
|
{
|
2015-02-13 02:01:24 +03:00
|
|
|
return running_clock() >> 30LL; /* 2^30 ~= 10^9 */
|
2010-05-08 01:11:44 +04:00
|
|
|
}
|
|
|
|
|
2012-12-18 03:59:50 +04:00
|
|
|
static void set_sample_period(void)
|
2010-05-08 01:11:44 +04:00
|
|
|
{
|
|
|
|
/*
|
2011-05-23 09:10:22 +04:00
|
|
|
* convert watchdog_thresh from seconds to ns
|
2012-02-10 02:42:22 +04:00
|
|
|
* the divide by 5 is to give hrtimer several chances (two
|
|
|
|
* or three with the current relation between the soft
|
|
|
|
* and hard thresholds) to increment before the
|
|
|
|
* hardlockup detector generates a warning
|
2010-05-08 01:11:44 +04:00
|
|
|
*/
|
2012-12-18 03:59:50 +04:00
|
|
|
sample_period = get_softlockup_thresh() * ((u64)NSEC_PER_SEC / 5);
|
2017-08-15 10:50:13 +03:00
|
|
|
watchdog_update_hrtimer_threshold(sample_period);
|
2010-05-08 01:11:44 +04:00
|
|
|
}
|
|
|
|
|
|
|
|
/* Commands for resetting the watchdog */
|
|
|
|
static void __touch_watchdog(void)
|
|
|
|
{
|
2012-12-27 06:49:44 +04:00
|
|
|
__this_cpu_write(watchdog_touch_ts, get_timestamp());
|
2010-05-08 01:11:44 +04:00
|
|
|
}
|
|
|
|
|
2015-12-08 19:28:04 +03:00
|
|
|
/**
|
|
|
|
* touch_softlockup_watchdog_sched - touch watchdog on scheduler stalls
|
|
|
|
*
|
|
|
|
* Call when the scheduler may have stalled for legitimate reasons
|
|
|
|
* preventing the watchdog task from executing - e.g. the scheduler
|
|
|
|
* entering idle state. This should only be used for scheduler events.
|
|
|
|
* Use touch_softlockup_watchdog() for everything else.
|
|
|
|
*/
|
|
|
|
void touch_softlockup_watchdog_sched(void)
|
2010-05-08 01:11:44 +04:00
|
|
|
{
|
2014-04-19 02:07:12 +04:00
|
|
|
/*
|
|
|
|
* Preemption can be enabled. It doesn't matter which CPU's timestamp
|
|
|
|
* gets zeroed here, so use the raw_ operation.
|
|
|
|
*/
|
|
|
|
raw_cpu_write(watchdog_touch_ts, 0);
|
2010-05-08 01:11:44 +04:00
|
|
|
}
|
2015-12-08 19:28:04 +03:00
|
|
|
|
|
|
|
void touch_softlockup_watchdog(void)
|
|
|
|
{
|
|
|
|
touch_softlockup_watchdog_sched();
|
workqueue: implement lockup detector
Workqueue stalls can happen from a variety of usage bugs such as
missing WQ_MEM_RECLAIM flag or concurrency managed work item
indefinitely staying RUNNING. These stalls can be extremely difficult
to hunt down because the usual warning mechanisms can't detect
workqueue stalls and the internal state is pretty opaque.
To alleviate the situation, this patch implements workqueue lockup
detector. It periodically monitors all worker_pools periodically and,
if any pool failed to make forward progress longer than the threshold
duration, triggers warning and dumps workqueue state as follows.
BUG: workqueue lockup - pool cpus=0 node=0 flags=0x0 nice=0 stuck for 31s!
Showing busy workqueues and worker pools:
workqueue events: flags=0x0
pwq 0: cpus=0 node=0 flags=0x0 nice=0 active=17/256
pending: monkey_wrench_fn, e1000_watchdog, cache_reap, vmstat_shepherd, release_one_tty, release_one_tty, release_one_tty, release_one_tty, release_one_tty, release_one_tty, release_one_tty, release_one_tty, release_one_tty, release_one_tty, release_one_tty, release_one_tty, cgroup_release_agent
workqueue events_power_efficient: flags=0x80
pwq 0: cpus=0 node=0 flags=0x0 nice=0 active=2/256
pending: check_lifetime, neigh_periodic_work
workqueue cgroup_pidlist_destroy: flags=0x0
pwq 0: cpus=0 node=0 flags=0x0 nice=0 active=1/1
pending: cgroup_pidlist_destroy_work_fn
...
The detection mechanism is controller through kernel parameter
workqueue.watchdog_thresh and can be updated at runtime through the
sysfs module parameter file.
v2: Decoupled from softlockup control knobs.
Signed-off-by: Tejun Heo <tj@kernel.org>
Acked-by: Don Zickus <dzickus@redhat.com>
Cc: Ulrich Obergfell <uobergfe@redhat.com>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Chris Mason <clm@fb.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
2015-12-08 19:28:04 +03:00
|
|
|
wq_watchdog_touch(raw_smp_processor_id());
|
2015-12-08 19:28:04 +03:00
|
|
|
}
|
2010-05-13 10:53:33 +04:00
|
|
|
EXPORT_SYMBOL(touch_softlockup_watchdog);
|
2010-05-08 01:11:44 +04:00
|
|
|
|
2010-05-08 01:11:45 +04:00
|
|
|
void touch_all_softlockup_watchdogs(void)
|
2010-05-08 01:11:44 +04:00
|
|
|
{
|
|
|
|
int cpu;
|
|
|
|
|
|
|
|
/*
|
2017-09-12 22:37:11 +03:00
|
|
|
* watchdog_mutex cannpt be taken here, as this might be called
|
|
|
|
* from (soft)interrupt context, so the access to
|
|
|
|
* watchdog_allowed_cpumask might race with a concurrent update.
|
|
|
|
*
|
|
|
|
* The watchdog time stamp can race against a concurrent real
|
|
|
|
* update as well, the only side effect might be a cycle delay for
|
|
|
|
* the softlockup check.
|
2010-05-08 01:11:44 +04:00
|
|
|
*/
|
2017-09-12 22:37:11 +03:00
|
|
|
for_each_cpu(cpu, &watchdog_allowed_mask)
|
2010-05-08 01:11:44 +04:00
|
|
|
per_cpu(watchdog_touch_ts, cpu) = 0;
|
workqueue: implement lockup detector
Workqueue stalls can happen from a variety of usage bugs such as
missing WQ_MEM_RECLAIM flag or concurrency managed work item
indefinitely staying RUNNING. These stalls can be extremely difficult
to hunt down because the usual warning mechanisms can't detect
workqueue stalls and the internal state is pretty opaque.
To alleviate the situation, this patch implements workqueue lockup
detector. It periodically monitors all worker_pools periodically and,
if any pool failed to make forward progress longer than the threshold
duration, triggers warning and dumps workqueue state as follows.
BUG: workqueue lockup - pool cpus=0 node=0 flags=0x0 nice=0 stuck for 31s!
Showing busy workqueues and worker pools:
workqueue events: flags=0x0
pwq 0: cpus=0 node=0 flags=0x0 nice=0 active=17/256
pending: monkey_wrench_fn, e1000_watchdog, cache_reap, vmstat_shepherd, release_one_tty, release_one_tty, release_one_tty, release_one_tty, release_one_tty, release_one_tty, release_one_tty, release_one_tty, release_one_tty, release_one_tty, release_one_tty, release_one_tty, cgroup_release_agent
workqueue events_power_efficient: flags=0x80
pwq 0: cpus=0 node=0 flags=0x0 nice=0 active=2/256
pending: check_lifetime, neigh_periodic_work
workqueue cgroup_pidlist_destroy: flags=0x0
pwq 0: cpus=0 node=0 flags=0x0 nice=0 active=1/1
pending: cgroup_pidlist_destroy_work_fn
...
The detection mechanism is controller through kernel parameter
workqueue.watchdog_thresh and can be updated at runtime through the
sysfs module parameter file.
v2: Decoupled from softlockup control knobs.
Signed-off-by: Tejun Heo <tj@kernel.org>
Acked-by: Don Zickus <dzickus@redhat.com>
Cc: Ulrich Obergfell <uobergfe@redhat.com>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Chris Mason <clm@fb.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
2015-12-08 19:28:04 +03:00
|
|
|
wq_watchdog_touch(-1);
|
2010-05-08 01:11:44 +04:00
|
|
|
}
|
|
|
|
|
|
|
|
void touch_softlockup_watchdog_sync(void)
|
|
|
|
{
|
2014-08-17 21:30:34 +04:00
|
|
|
__this_cpu_write(softlockup_touch_sync, true);
|
|
|
|
__this_cpu_write(watchdog_touch_ts, 0);
|
2010-05-08 01:11:44 +04:00
|
|
|
}
|
|
|
|
|
2010-05-18 02:06:04 +04:00
|
|
|
static int is_softlockup(unsigned long touch_ts)
|
2010-05-08 01:11:44 +04:00
|
|
|
{
|
2012-12-27 06:49:44 +04:00
|
|
|
unsigned long now = get_timestamp();
|
2010-05-08 01:11:44 +04:00
|
|
|
|
2015-11-06 05:44:56 +03:00
|
|
|
if ((watchdog_enabled & SOFT_WATCHDOG_ENABLED) && watchdog_thresh){
|
2015-04-15 01:44:13 +03:00
|
|
|
/* Warn about unreasonable delays. */
|
|
|
|
if (time_after(now, touch_ts + get_softlockup_thresh()))
|
|
|
|
return now - touch_ts;
|
|
|
|
}
|
2010-05-08 01:11:44 +04:00
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2017-07-13 00:35:46 +03:00
|
|
|
/* watchdog detector functions */
|
|
|
|
bool is_hardlockup(void)
|
2010-05-08 01:11:44 +04:00
|
|
|
{
|
2017-07-13 00:35:46 +03:00
|
|
|
unsigned long hrint = __this_cpu_read(hrtimer_interrupts);
|
2012-07-16 14:42:38 +04:00
|
|
|
|
2017-07-13 00:35:46 +03:00
|
|
|
if (__this_cpu_read(hrtimer_interrupts_saved) == hrint)
|
|
|
|
return true;
|
|
|
|
|
|
|
|
__this_cpu_write(hrtimer_interrupts_saved, hrint);
|
|
|
|
return false;
|
2016-12-15 02:06:24 +03:00
|
|
|
}
|
2017-07-13 00:35:46 +03:00
|
|
|
|
|
|
|
static void watchdog_interrupt_count(void)
|
2016-12-15 02:06:24 +03:00
|
|
|
{
|
2017-07-13 00:35:46 +03:00
|
|
|
__this_cpu_inc(hrtimer_interrupts);
|
2016-12-15 02:06:24 +03:00
|
|
|
}
|
2010-05-08 01:11:44 +04:00
|
|
|
|
|
|
|
/* watchdog kicker functions */
|
|
|
|
static enum hrtimer_restart watchdog_timer_fn(struct hrtimer *hrtimer)
|
|
|
|
{
|
2010-12-08 18:22:55 +03:00
|
|
|
unsigned long touch_ts = __this_cpu_read(watchdog_touch_ts);
|
2010-05-08 01:11:44 +04:00
|
|
|
struct pt_regs *regs = get_irq_regs();
|
|
|
|
int duration;
|
2014-06-24 00:22:05 +04:00
|
|
|
int softlockup_all_cpu_backtrace = sysctl_softlockup_all_cpu_backtrace;
|
2010-05-08 01:11:44 +04:00
|
|
|
|
2017-09-12 22:37:05 +03:00
|
|
|
if (!watchdog_enabled)
|
2017-01-25 02:17:53 +03:00
|
|
|
return HRTIMER_NORESTART;
|
|
|
|
|
2010-05-08 01:11:44 +04:00
|
|
|
/* kick the hardlockup detector */
|
|
|
|
watchdog_interrupt_count();
|
|
|
|
|
|
|
|
/* kick the softlockup detector */
|
2010-12-08 18:22:55 +03:00
|
|
|
wake_up_process(__this_cpu_read(softlockup_watchdog));
|
2010-05-08 01:11:44 +04:00
|
|
|
|
|
|
|
/* .. and repeat */
|
2012-12-18 03:59:50 +04:00
|
|
|
hrtimer_forward_now(hrtimer, ns_to_ktime(sample_period));
|
2010-05-08 01:11:44 +04:00
|
|
|
|
|
|
|
if (touch_ts == 0) {
|
2010-12-08 18:22:55 +03:00
|
|
|
if (unlikely(__this_cpu_read(softlockup_touch_sync))) {
|
2010-05-08 01:11:44 +04:00
|
|
|
/*
|
|
|
|
* If the time stamp was touched atomically
|
|
|
|
* make sure the scheduler tick is up to date.
|
|
|
|
*/
|
2010-12-08 18:22:55 +03:00
|
|
|
__this_cpu_write(softlockup_touch_sync, false);
|
2010-05-08 01:11:44 +04:00
|
|
|
sched_clock_tick();
|
|
|
|
}
|
2012-03-10 23:37:28 +04:00
|
|
|
|
|
|
|
/* Clear the guest paused flag on watchdog reset */
|
|
|
|
kvm_check_and_clear_guest_paused();
|
2010-05-08 01:11:44 +04:00
|
|
|
__touch_watchdog();
|
|
|
|
return HRTIMER_RESTART;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* check for a softlockup
|
|
|
|
* This is done by making sure a high priority task is
|
|
|
|
* being scheduled. The task touches the watchdog to
|
|
|
|
* indicate it is getting cpu time. If it hasn't then
|
|
|
|
* this is a good indication some task is hogging the cpu
|
|
|
|
*/
|
2010-05-18 02:06:04 +04:00
|
|
|
duration = is_softlockup(touch_ts);
|
2010-05-08 01:11:44 +04:00
|
|
|
if (unlikely(duration)) {
|
2012-03-10 23:37:28 +04:00
|
|
|
/*
|
|
|
|
* If a virtual machine is stopped by the host it can look to
|
|
|
|
* the watchdog like a soft lockup, check to see if the host
|
|
|
|
* stopped the vm before we issue the warning
|
|
|
|
*/
|
|
|
|
if (kvm_check_and_clear_guest_paused())
|
|
|
|
return HRTIMER_RESTART;
|
|
|
|
|
2010-05-08 01:11:44 +04:00
|
|
|
/* only warn once */
|
2014-10-10 02:25:17 +04:00
|
|
|
if (__this_cpu_read(soft_watchdog_warn) == true) {
|
|
|
|
/*
|
|
|
|
* When multiple processes are causing softlockups the
|
|
|
|
* softlockup detector only warns on the first one
|
|
|
|
* because the code relies on a full quiet cycle to
|
|
|
|
* re-arm. The second process prevents the quiet cycle
|
|
|
|
* and never gets reported. Use task pointers to detect
|
|
|
|
* this.
|
|
|
|
*/
|
|
|
|
if (__this_cpu_read(softlockup_task_ptr_saved) !=
|
|
|
|
current) {
|
|
|
|
__this_cpu_write(soft_watchdog_warn, false);
|
|
|
|
__touch_watchdog();
|
|
|
|
}
|
2010-05-08 01:11:44 +04:00
|
|
|
return HRTIMER_RESTART;
|
2014-10-10 02:25:17 +04:00
|
|
|
}
|
2010-05-08 01:11:44 +04:00
|
|
|
|
2014-06-24 00:22:05 +04:00
|
|
|
if (softlockup_all_cpu_backtrace) {
|
|
|
|
/* Prevent multiple soft-lockup reports if one cpu is already
|
|
|
|
* engaged in dumping cpu back traces
|
|
|
|
*/
|
|
|
|
if (test_and_set_bit(0, &soft_lockup_nmi_warn)) {
|
|
|
|
/* Someone else will report us. Let's give up */
|
|
|
|
__this_cpu_write(soft_watchdog_warn, true);
|
|
|
|
return HRTIMER_RESTART;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2014-08-07 03:04:03 +04:00
|
|
|
pr_emerg("BUG: soft lockup - CPU#%d stuck for %us! [%s:%d]\n",
|
2010-05-18 02:06:04 +04:00
|
|
|
smp_processor_id(), duration,
|
2010-05-08 01:11:44 +04:00
|
|
|
current->comm, task_pid_nr(current));
|
2014-10-10 02:25:17 +04:00
|
|
|
__this_cpu_write(softlockup_task_ptr_saved, current);
|
2010-05-08 01:11:44 +04:00
|
|
|
print_modules();
|
|
|
|
print_irqtrace_events(current);
|
|
|
|
if (regs)
|
|
|
|
show_regs(regs);
|
|
|
|
else
|
|
|
|
dump_stack();
|
|
|
|
|
2014-06-24 00:22:05 +04:00
|
|
|
if (softlockup_all_cpu_backtrace) {
|
|
|
|
/* Avoid generating two back traces for current
|
|
|
|
* given that one is already made above
|
|
|
|
*/
|
|
|
|
trigger_allbutself_cpu_backtrace();
|
|
|
|
|
|
|
|
clear_bit(0, &soft_lockup_nmi_warn);
|
|
|
|
/* Barrier to sync with other cpus */
|
|
|
|
smp_mb__after_atomic();
|
|
|
|
}
|
|
|
|
|
2014-08-09 01:22:31 +04:00
|
|
|
add_taint(TAINT_SOFTLOCKUP, LOCKDEP_STILL_OK);
|
2010-05-08 01:11:44 +04:00
|
|
|
if (softlockup_panic)
|
|
|
|
panic("softlockup: hung tasks");
|
2010-12-08 18:22:55 +03:00
|
|
|
__this_cpu_write(soft_watchdog_warn, true);
|
2010-05-08 01:11:44 +04:00
|
|
|
} else
|
2010-12-08 18:22:55 +03:00
|
|
|
__this_cpu_write(soft_watchdog_warn, false);
|
2010-05-08 01:11:44 +04:00
|
|
|
|
|
|
|
return HRTIMER_RESTART;
|
|
|
|
}
|
|
|
|
|
2012-07-16 14:42:38 +04:00
|
|
|
static void watchdog_set_prio(unsigned int policy, unsigned int prio)
|
|
|
|
{
|
|
|
|
struct sched_param param = { .sched_priority = prio };
|
2010-05-08 01:11:44 +04:00
|
|
|
|
2012-07-16 14:42:38 +04:00
|
|
|
sched_setscheduler(current, policy, ¶m);
|
|
|
|
}
|
|
|
|
|
|
|
|
static void watchdog_enable(unsigned int cpu)
|
2010-05-08 01:11:44 +04:00
|
|
|
{
|
2017-09-12 22:37:05 +03:00
|
|
|
struct hrtimer *hrtimer = this_cpu_ptr(&watchdog_hrtimer);
|
2010-05-08 01:11:44 +04:00
|
|
|
|
2017-09-12 22:37:05 +03:00
|
|
|
/*
|
|
|
|
* Start the timer first to prevent the NMI watchdog triggering
|
|
|
|
* before the timer has a chance to fire.
|
|
|
|
*/
|
2012-12-19 23:51:31 +04:00
|
|
|
hrtimer_init(hrtimer, CLOCK_MONOTONIC, HRTIMER_MODE_REL);
|
|
|
|
hrtimer->function = watchdog_timer_fn;
|
2017-09-12 22:37:05 +03:00
|
|
|
hrtimer_start(hrtimer, ns_to_ktime(sample_period),
|
|
|
|
HRTIMER_MODE_REL_PINNED);
|
2012-12-19 23:51:31 +04:00
|
|
|
|
2017-09-12 22:37:05 +03:00
|
|
|
/* Initialize timestamp */
|
|
|
|
__touch_watchdog();
|
2012-07-16 14:42:38 +04:00
|
|
|
/* Enable the perf event */
|
|
|
|
watchdog_nmi_enable(cpu);
|
2010-05-08 01:11:44 +04:00
|
|
|
|
2012-07-16 14:42:38 +04:00
|
|
|
watchdog_set_prio(SCHED_FIFO, MAX_RT_PRIO - 1);
|
|
|
|
}
|
2010-05-08 01:11:44 +04:00
|
|
|
|
2012-07-16 14:42:38 +04:00
|
|
|
static void watchdog_disable(unsigned int cpu)
|
|
|
|
{
|
2017-09-12 22:37:05 +03:00
|
|
|
struct hrtimer *hrtimer = this_cpu_ptr(&watchdog_hrtimer);
|
2010-05-08 01:11:44 +04:00
|
|
|
|
2012-07-16 14:42:38 +04:00
|
|
|
watchdog_set_prio(SCHED_NORMAL, 0);
|
2017-09-12 22:37:05 +03:00
|
|
|
/*
|
|
|
|
* Disable the perf event first. That prevents that a large delay
|
|
|
|
* between disabling the timer and disabling the perf event causes
|
|
|
|
* the perf NMI to detect a false positive.
|
|
|
|
*/
|
2012-07-16 14:42:38 +04:00
|
|
|
watchdog_nmi_disable(cpu);
|
2017-09-12 22:37:05 +03:00
|
|
|
hrtimer_cancel(hrtimer);
|
2010-05-08 01:11:44 +04:00
|
|
|
}
|
|
|
|
|
2013-06-06 17:42:53 +04:00
|
|
|
static void watchdog_cleanup(unsigned int cpu, bool online)
|
|
|
|
{
|
|
|
|
watchdog_disable(cpu);
|
|
|
|
}
|
|
|
|
|
2012-07-16 14:42:38 +04:00
|
|
|
static int watchdog_should_run(unsigned int cpu)
|
|
|
|
{
|
|
|
|
return __this_cpu_read(hrtimer_interrupts) !=
|
|
|
|
__this_cpu_read(soft_lockup_hrtimer_cnt);
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* The watchdog thread function - touches the timestamp.
|
|
|
|
*
|
2012-12-18 03:59:50 +04:00
|
|
|
* It only runs once every sample_period seconds (4 seconds by
|
2012-07-16 14:42:38 +04:00
|
|
|
* default) to reset the softlockup timestamp. If this gets delayed
|
|
|
|
* for more than 2*watchdog_thresh seconds then the debug-printout
|
|
|
|
* triggers in watchdog_timer_fn().
|
|
|
|
*/
|
|
|
|
static void watchdog(unsigned int cpu)
|
|
|
|
{
|
|
|
|
__this_cpu_write(soft_lockup_hrtimer_cnt,
|
|
|
|
__this_cpu_read(hrtimer_interrupts));
|
|
|
|
__touch_watchdog();
|
|
|
|
}
|
2010-05-08 01:11:44 +04:00
|
|
|
|
2013-06-06 17:42:53 +04:00
|
|
|
static struct smp_hotplug_thread watchdog_threads = {
|
|
|
|
.store = &softlockup_watchdog,
|
|
|
|
.thread_should_run = watchdog_should_run,
|
|
|
|
.thread_fn = watchdog,
|
|
|
|
.thread_comm = "watchdog/%u",
|
|
|
|
.setup = watchdog_enable,
|
|
|
|
.cleanup = watchdog_cleanup,
|
|
|
|
.park = watchdog_disable,
|
|
|
|
.unpark = watchdog_enable,
|
|
|
|
};
|
|
|
|
|
2017-09-12 22:37:10 +03:00
|
|
|
static void softlockup_update_smpboot_threads(void)
|
|
|
|
{
|
|
|
|
lockdep_assert_held(&watchdog_mutex);
|
|
|
|
|
|
|
|
if (!softlockup_threads_initialized)
|
|
|
|
return;
|
|
|
|
|
|
|
|
smpboot_update_cpumask_percpu_thread(&watchdog_threads,
|
|
|
|
&watchdog_allowed_mask);
|
|
|
|
__lockup_detector_cleanup();
|
|
|
|
}
|
|
|
|
|
|
|
|
/* Temporarily park all watchdog threads */
|
|
|
|
static void softlockup_park_all_threads(void)
|
|
|
|
{
|
|
|
|
cpumask_clear(&watchdog_allowed_mask);
|
|
|
|
softlockup_update_smpboot_threads();
|
|
|
|
}
|
|
|
|
|
2017-09-12 22:37:12 +03:00
|
|
|
/* Unpark enabled threads */
|
|
|
|
static void softlockup_unpark_threads(void)
|
2017-09-12 22:37:10 +03:00
|
|
|
{
|
|
|
|
cpumask_copy(&watchdog_allowed_mask, &watchdog_cpumask);
|
|
|
|
softlockup_update_smpboot_threads();
|
|
|
|
}
|
|
|
|
|
2017-09-12 22:37:17 +03:00
|
|
|
static void softlockup_reconfigure_threads(void)
|
2017-09-12 22:37:10 +03:00
|
|
|
{
|
2017-09-12 22:37:16 +03:00
|
|
|
watchdog_nmi_reconfigure(false);
|
2017-09-12 22:37:10 +03:00
|
|
|
softlockup_park_all_threads();
|
|
|
|
set_sample_period();
|
2017-09-12 22:37:17 +03:00
|
|
|
lockup_detector_update_enable();
|
|
|
|
if (watchdog_enabled && watchdog_thresh)
|
2017-09-12 22:37:12 +03:00
|
|
|
softlockup_unpark_threads();
|
2017-09-12 22:37:16 +03:00
|
|
|
watchdog_nmi_reconfigure(true);
|
2017-09-12 22:37:10 +03:00
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Create the watchdog thread infrastructure.
|
|
|
|
*
|
|
|
|
* The threads are not unparked as watchdog_allowed_mask is empty. When
|
|
|
|
* the threads are sucessfully initialized, take the proper locks and
|
|
|
|
* unpark the threads in the watchdog_cpumask if the watchdog is enabled.
|
|
|
|
*/
|
|
|
|
static __init void softlockup_init_threads(void)
|
|
|
|
{
|
|
|
|
int ret;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* If sysctl is off and watchdog got disabled on the command line,
|
|
|
|
* nothing to do here.
|
|
|
|
*/
|
2017-09-12 22:37:17 +03:00
|
|
|
lockup_detector_update_enable();
|
|
|
|
|
2017-09-12 22:37:10 +03:00
|
|
|
if (!IS_ENABLED(CONFIG_SYSCTL) &&
|
|
|
|
!(watchdog_enabled && watchdog_thresh))
|
|
|
|
return;
|
|
|
|
|
|
|
|
ret = smpboot_register_percpu_thread_cpumask(&watchdog_threads,
|
|
|
|
&watchdog_allowed_mask);
|
|
|
|
if (ret) {
|
|
|
|
pr_err("Failed to initialize soft lockup detector threads\n");
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
|
|
|
mutex_lock(&watchdog_mutex);
|
|
|
|
softlockup_threads_initialized = true;
|
2017-09-12 22:37:17 +03:00
|
|
|
softlockup_reconfigure_threads();
|
2017-09-12 22:37:10 +03:00
|
|
|
mutex_unlock(&watchdog_mutex);
|
|
|
|
}
|
|
|
|
|
2017-09-12 22:37:06 +03:00
|
|
|
#else /* CONFIG_SOFTLOCKUP_DETECTOR */
|
|
|
|
static inline int watchdog_park_threads(void) { return 0; }
|
|
|
|
static inline void watchdog_unpark_threads(void) { }
|
|
|
|
static inline int watchdog_enable_all_cpus(void) { return 0; }
|
|
|
|
static inline void watchdog_disable_all_cpus(void) { }
|
2017-09-12 22:37:10 +03:00
|
|
|
static inline void softlockup_init_threads(void) { }
|
2017-09-12 22:37:17 +03:00
|
|
|
static void softlockup_reconfigure_threads(void)
|
2017-09-12 22:37:16 +03:00
|
|
|
{
|
|
|
|
watchdog_nmi_reconfigure(false);
|
2017-09-12 22:37:17 +03:00
|
|
|
lockup_detector_update_enable();
|
2017-09-12 22:37:16 +03:00
|
|
|
watchdog_nmi_reconfigure(true);
|
|
|
|
}
|
2017-09-12 22:37:06 +03:00
|
|
|
#endif /* !CONFIG_SOFTLOCKUP_DETECTOR */
|
2017-07-13 00:35:46 +03:00
|
|
|
|
2017-09-12 22:37:04 +03:00
|
|
|
static void __lockup_detector_cleanup(void)
|
|
|
|
{
|
|
|
|
lockdep_assert_held(&watchdog_mutex);
|
|
|
|
hardlockup_detector_perf_cleanup();
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* lockup_detector_cleanup - Cleanup after cpu hotplug or sysctl changes
|
|
|
|
*
|
|
|
|
* Caller must not hold the cpu hotplug rwsem.
|
|
|
|
*/
|
|
|
|
void lockup_detector_cleanup(void)
|
|
|
|
{
|
|
|
|
mutex_lock(&watchdog_mutex);
|
|
|
|
__lockup_detector_cleanup();
|
|
|
|
mutex_unlock(&watchdog_mutex);
|
|
|
|
}
|
|
|
|
|
2017-09-12 22:36:57 +03:00
|
|
|
/**
|
|
|
|
* lockup_detector_soft_poweroff - Interface to stop lockup detector(s)
|
|
|
|
*
|
|
|
|
* Special interface for parisc. It prevents lockup detector warnings from
|
|
|
|
* the default pm_poweroff() function which busy loops forever.
|
|
|
|
*/
|
|
|
|
void lockup_detector_soft_poweroff(void)
|
|
|
|
{
|
|
|
|
watchdog_enabled = 0;
|
|
|
|
}
|
|
|
|
|
2015-11-06 05:44:30 +03:00
|
|
|
#ifdef CONFIG_SYSCTL
|
|
|
|
|
2017-09-12 22:37:12 +03:00
|
|
|
/* Propagate any changes to the watchdog threads */
|
2017-09-12 22:37:11 +03:00
|
|
|
static void proc_watchdog_update(void)
|
2015-04-15 01:43:58 +03:00
|
|
|
{
|
2017-09-12 22:37:12 +03:00
|
|
|
/* Remove impossible cpus to keep sysctl output clean. */
|
|
|
|
cpumask_and(&watchdog_cpumask, &watchdog_cpumask, cpu_possible_mask);
|
2017-09-12 22:37:17 +03:00
|
|
|
softlockup_reconfigure_threads();
|
2015-04-15 01:43:58 +03:00
|
|
|
}
|
|
|
|
|
2015-04-15 01:44:05 +03:00
|
|
|
/*
|
|
|
|
* common function for watchdog, nmi_watchdog and soft_watchdog parameter
|
|
|
|
*
|
2017-09-12 22:37:15 +03:00
|
|
|
* caller | table->data points to | 'which'
|
|
|
|
* -------------------|----------------------------|--------------------------
|
|
|
|
* proc_watchdog | watchdog_user_enabled | NMI_WATCHDOG_ENABLED |
|
|
|
|
* | | SOFT_WATCHDOG_ENABLED
|
|
|
|
* -------------------|----------------------------|--------------------------
|
|
|
|
* proc_nmi_watchdog | nmi_watchdog_user_enabled | NMI_WATCHDOG_ENABLED
|
|
|
|
* -------------------|----------------------------|--------------------------
|
|
|
|
* proc_soft_watchdog | soft_watchdog_user_enabled | SOFT_WATCHDOG_ENABLED
|
2015-04-15 01:44:05 +03:00
|
|
|
*/
|
|
|
|
static int proc_watchdog_common(int which, struct ctl_table *table, int write,
|
|
|
|
void __user *buffer, size_t *lenp, loff_t *ppos)
|
|
|
|
{
|
2017-09-12 22:37:17 +03:00
|
|
|
int err, old, *param = table->data;
|
2015-04-15 01:44:05 +03:00
|
|
|
|
2017-09-12 22:37:00 +03:00
|
|
|
cpu_hotplug_disable();
|
2017-09-12 22:37:01 +03:00
|
|
|
mutex_lock(&watchdog_mutex);
|
2015-04-15 01:44:05 +03:00
|
|
|
|
|
|
|
if (!write) {
|
2017-09-12 22:37:17 +03:00
|
|
|
/*
|
|
|
|
* On read synchronize the userspace interface. This is a
|
|
|
|
* racy snapshot.
|
|
|
|
*/
|
|
|
|
*param = (watchdog_enabled & which) != 0;
|
2015-04-15 01:44:05 +03:00
|
|
|
err = proc_dointvec_minmax(table, write, buffer, lenp, ppos);
|
|
|
|
} else {
|
2017-09-12 22:37:17 +03:00
|
|
|
old = READ_ONCE(*param);
|
2015-04-15 01:44:05 +03:00
|
|
|
err = proc_dointvec_minmax(table, write, buffer, lenp, ppos);
|
2017-09-12 22:37:17 +03:00
|
|
|
if (!err && old != READ_ONCE(*param))
|
2017-09-12 22:37:11 +03:00
|
|
|
proc_watchdog_update();
|
2015-04-15 01:44:05 +03:00
|
|
|
}
|
2017-09-12 22:37:01 +03:00
|
|
|
mutex_unlock(&watchdog_mutex);
|
2017-09-12 22:37:00 +03:00
|
|
|
cpu_hotplug_enable();
|
2015-04-15 01:44:05 +03:00
|
|
|
return err;
|
|
|
|
}
|
|
|
|
|
2015-04-15 01:44:08 +03:00
|
|
|
/*
|
|
|
|
* /proc/sys/kernel/watchdog
|
|
|
|
*/
|
|
|
|
int proc_watchdog(struct ctl_table *table, int write,
|
|
|
|
void __user *buffer, size_t *lenp, loff_t *ppos)
|
|
|
|
{
|
|
|
|
return proc_watchdog_common(NMI_WATCHDOG_ENABLED|SOFT_WATCHDOG_ENABLED,
|
|
|
|
table, write, buffer, lenp, ppos);
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* /proc/sys/kernel/nmi_watchdog
|
2010-05-08 01:11:44 +04:00
|
|
|
*/
|
2015-04-15 01:44:08 +03:00
|
|
|
int proc_nmi_watchdog(struct ctl_table *table, int write,
|
|
|
|
void __user *buffer, size_t *lenp, loff_t *ppos)
|
|
|
|
{
|
2017-09-12 22:37:19 +03:00
|
|
|
if (!nmi_watchdog_available && write)
|
|
|
|
return -ENOTSUPP;
|
2015-04-15 01:44:08 +03:00
|
|
|
return proc_watchdog_common(NMI_WATCHDOG_ENABLED,
|
|
|
|
table, write, buffer, lenp, ppos);
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* /proc/sys/kernel/soft_watchdog
|
|
|
|
*/
|
|
|
|
int proc_soft_watchdog(struct ctl_table *table, int write,
|
|
|
|
void __user *buffer, size_t *lenp, loff_t *ppos)
|
|
|
|
{
|
|
|
|
return proc_watchdog_common(SOFT_WATCHDOG_ENABLED,
|
|
|
|
table, write, buffer, lenp, ppos);
|
|
|
|
}
|
2010-05-08 01:11:44 +04:00
|
|
|
|
2015-04-15 01:44:08 +03:00
|
|
|
/*
|
|
|
|
* /proc/sys/kernel/watchdog_thresh
|
|
|
|
*/
|
|
|
|
int proc_watchdog_thresh(struct ctl_table *table, int write,
|
|
|
|
void __user *buffer, size_t *lenp, loff_t *ppos)
|
2010-05-08 01:11:44 +04:00
|
|
|
{
|
2017-09-12 22:37:11 +03:00
|
|
|
int err, old;
|
2010-05-08 01:11:44 +04:00
|
|
|
|
2017-09-12 22:37:00 +03:00
|
|
|
cpu_hotplug_disable();
|
2017-09-12 22:37:01 +03:00
|
|
|
mutex_lock(&watchdog_mutex);
|
2012-07-16 14:42:38 +04:00
|
|
|
|
2017-09-12 22:37:11 +03:00
|
|
|
old = READ_ONCE(watchdog_thresh);
|
2013-06-06 17:42:53 +04:00
|
|
|
err = proc_dointvec_minmax(table, write, buffer, lenp, ppos);
|
2015-04-15 01:44:08 +03:00
|
|
|
|
2017-09-12 22:37:11 +03:00
|
|
|
if (!err && write && old != READ_ONCE(watchdog_thresh))
|
|
|
|
proc_watchdog_update();
|
2011-05-23 09:10:21 +04:00
|
|
|
|
2017-09-12 22:37:01 +03:00
|
|
|
mutex_unlock(&watchdog_mutex);
|
2017-09-12 22:37:00 +03:00
|
|
|
cpu_hotplug_enable();
|
2013-06-06 17:42:53 +04:00
|
|
|
return err;
|
2010-05-08 01:11:44 +04:00
|
|
|
}
|
2015-06-25 02:55:45 +03:00
|
|
|
|
|
|
|
/*
|
|
|
|
* The cpumask is the mask of possible cpus that the watchdog can run
|
|
|
|
* on, not the mask of cpus it is actually running on. This allows the
|
|
|
|
* user to specify a mask that will include cpus that have not yet
|
|
|
|
* been brought online, if desired.
|
|
|
|
*/
|
|
|
|
int proc_watchdog_cpumask(struct ctl_table *table, int write,
|
|
|
|
void __user *buffer, size_t *lenp, loff_t *ppos)
|
|
|
|
{
|
|
|
|
int err;
|
|
|
|
|
2017-09-12 22:37:00 +03:00
|
|
|
cpu_hotplug_disable();
|
2017-09-12 22:37:01 +03:00
|
|
|
mutex_lock(&watchdog_mutex);
|
2015-09-05 01:45:18 +03:00
|
|
|
|
2015-06-25 02:55:45 +03:00
|
|
|
err = proc_do_large_bitmap(table, write, buffer, lenp, ppos);
|
2017-09-12 22:37:08 +03:00
|
|
|
if (!err && write)
|
2017-09-12 22:37:12 +03:00
|
|
|
proc_watchdog_update();
|
2017-09-12 22:36:59 +03:00
|
|
|
|
2017-09-12 22:37:01 +03:00
|
|
|
mutex_unlock(&watchdog_mutex);
|
2017-09-12 22:37:00 +03:00
|
|
|
cpu_hotplug_enable();
|
2015-06-25 02:55:45 +03:00
|
|
|
return err;
|
|
|
|
}
|
2010-05-08 01:11:44 +04:00
|
|
|
#endif /* CONFIG_SYSCTL */
|
|
|
|
|
2010-11-25 20:38:29 +03:00
|
|
|
void __init lockup_detector_init(void)
|
2010-05-08 01:11:44 +04:00
|
|
|
{
|
2015-06-25 02:55:45 +03:00
|
|
|
#ifdef CONFIG_NO_HZ_FULL
|
|
|
|
if (tick_nohz_full_enabled()) {
|
2015-09-05 01:45:09 +03:00
|
|
|
pr_info("Disabling watchdog on nohz_full cores by default\n");
|
|
|
|
cpumask_copy(&watchdog_cpumask, housekeeping_mask);
|
2015-06-25 02:55:45 +03:00
|
|
|
} else
|
|
|
|
cpumask_copy(&watchdog_cpumask, cpu_possible_mask);
|
|
|
|
#else
|
|
|
|
cpumask_copy(&watchdog_cpumask, cpu_possible_mask);
|
|
|
|
#endif
|
|
|
|
|
2017-09-12 22:37:19 +03:00
|
|
|
if (!watchdog_nmi_probe())
|
|
|
|
nmi_watchdog_available = true;
|
2017-09-12 22:37:11 +03:00
|
|
|
softlockup_init_threads();
|
2010-05-08 01:11:44 +04:00
|
|
|
}
|