@@ -40,6 +40,7 @@ show up in /proc/sys/kernel:
- hotplug
- hardlockup_all_cpu_backtrace
- hardlockup_panic
+- hung_task_all_cpu_backtrace
- hung_task_panic
- hung_task_check_count
- hung_task_timeout_secs
@@ -338,6 +339,20 @@ Path for the hotplug policy agent.
Default value is "/sbin/hotplug".
+hung_task_all_cpu_backtrace:
+================
+
+If this option is set, the kernel will send an NMI to all CPUs to dump
+their backtraces when a hung task is detected. This file shows up if
+CONFIG_DETECT_HUNG_TASK and CONFIG_SMP are enabled.
+
+0: Won't show all CPUs backtraces when a hung task is detected.
+This is the default behavior.
+
+1: Will non-maskably interrupt all CPUs and dump their backtraces when
+a hung task is detected.
+
+
hung_task_panic:
================
@@ -7,6 +7,13 @@
struct ctl_table;
#ifdef CONFIG_DETECT_HUNG_TASK
+
+#ifdef CONFIG_SMP
+extern unsigned int sysctl_hung_task_all_cpu_backtrace;
+#else
+#define sysctl_hung_task_all_cpu_backtrace 0
+#endif /* CONFIG_SMP */
+
extern int sysctl_hung_task_check_count;
extern unsigned int sysctl_hung_task_panic;
extern unsigned long sysctl_hung_task_timeout_secs;
@@ -53,9 +53,18 @@ int __read_mostly sysctl_hung_task_warnings = 10;
static int __read_mostly did_panic;
static bool hung_task_show_lock;
static bool hung_task_call_panic;
+static bool hung_task_show_all_bt;
static struct task_struct *watchdog_task;
+#ifdef CONFIG_SMP
+/*
+ * Should we dump all CPUs backtraces in a hung task event?
+ * Defaults to 0, can be changed via sysctl.
+ */
+unsigned int __read_mostly sysctl_hung_task_all_cpu_backtrace;
+#endif /* CONFIG_SMP */
+
/*
* Should we panic (and reboot, if panic_timeout= is set) when a
* hung task is detected:
@@ -137,6 +146,9 @@ static void check_hung_task(struct task_struct *t, unsigned long timeout)
" disables this message.\n");
sched_show_task(t);
hung_task_show_lock = true;
+
+ if (sysctl_hung_task_all_cpu_backtrace)
+ hung_task_show_all_bt = true;
}
touch_nmi_watchdog();
@@ -201,10 +213,14 @@ static void check_hung_uninterruptible_tasks(unsigned long timeout)
rcu_read_unlock();
if (hung_task_show_lock)
debug_show_all_locks();
- if (hung_task_call_panic) {
+
+ if (hung_task_show_all_bt) {
+ hung_task_show_all_bt = false;
trigger_all_cpu_backtrace();
+ }
+
+ if (hung_task_call_panic)
panic("hung_task: blocked tasks");
- }
}
static long hung_timeout_jiffies(unsigned long last_checked,
@@ -1098,6 +1098,17 @@ static struct ctl_table kern_table[] = {
},
#endif
#ifdef CONFIG_DETECT_HUNG_TASK
+#ifdef CONFIG_SMP
+ {
+ .procname = "hung_task_all_cpu_backtrace",
+ .data = &sysctl_hung_task_all_cpu_backtrace,
+ .maxlen = sizeof(int),
+ .mode = 0644,
+ .proc_handler = proc_dointvec_minmax,
+ .extra1 = SYSCTL_ZERO,
+ .extra2 = SYSCTL_ONE,
+ },
+#endif /* CONFIG_SMP */
{
.procname = "hung_task_panic",
.data = &sysctl_hung_task_panic,
Commit 401c636a0eeb ("kernel/hung_task.c: show all hung tasks before panic") introduced a change in that we started to show all CPUs backtraces when a hung task is detected _and_ the sysctl/kernel parameter "hung_task_panic" is set. The idea is good, because usually when observing deadlocks (that may lead to hung tasks), the culprit is another task holding a lock and not necessarily the task detected as hung. The problem with this approach is that dumping backtraces is a slightly expensive task, specially printing that on console (and specially in many CPU machines, as servers commonly found nowadays). So, users that plan to collect a kdump to investigate the hung tasks and narrow down the deadlock definitely don't need the CPUs backtrace on dmesg/console, which will delay the panic and pollute the log (crash tool would easily grab all CPUs traces with 'bt -a' command). Also, there's the reciprocal scenario: some users may be interested in seeing the CPUs backtraces but not have the system panic when a hung task is detected. The current approach hence is almost as embedding a policy in the kernel, by forcing the CPUs backtraces' dump (only) on hung_task_panic. This patch decouples the panic event on hung task from the CPUs backtraces dump, by creating (and documenting) a new sysctl called "hung_task_all_cpu_backtrace", analog to the approach taken on soft/hard lockups, that have both a panic and an "all_cpu_backtrace" sysctl to allow individual control. The new mechanism for dumping the CPUs backtraces on hung task detection respects "hung_task_warnings" by not dumping the traces in case there's no warnings left. Cc: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp> Reviewed-by: Kees Cook <keescook@chromium.org> Signed-off-by: Guilherme G. Piccoli <gpiccoli@canonical.com> --- V3: Following the suggestion from Vlastimil, removed the kernel parameter since soon we (hopefully) will have a generic approach to set sysctls via kernel parameters[0] - thanks Vlastimil, great idea! Thanks, Guilherme [0] lore.kernel.org/lkml/20200326181606.7027-1-vbabka@suse.cz/T Documentation/admin-guide/sysctl/kernel.rst | 15 +++++++++++++++ include/linux/sched/sysctl.h | 7 +++++++ kernel/hung_task.c | 20 ++++++++++++++++++-- kernel/sysctl.c | 11 +++++++++++ 4 files changed, 51 insertions(+), 2 deletions(-)