Some kernel threads are always in D state, when we enable hung_task,
it will misjudge, we should skip these to narrow the scope.
exp mtk mobilephone:
root 420 420 2 0 0 kwdt_thread 0 D wdtk-0
root 421 421 2 0 0 kwdt_thread 0 D wdtk-1
root 422 422 2 0 0 kwdt_thread 0 D wdtk-2
root 423 423 2 0 0 kwdt_thread 0 D wdtk-3
root 424 424 2 0 0 kwdt_thread 0 D wdtk-4
root 425 425 2 0 0 kwdt_thread 0 D wdtk-5
root 426 426 2 0 0 kwdt_thread 0 D wdtk-6
root 427 427 2 0 0 kwdt_thread 0 D wdtk-7
root 435 435 2 0 0 mtk_lpm_monitor_thread 0 D LPM-0
root 436 436 2 0 0 mtk_lpm_monitor_thread 0 D LPM-1
root 437 437 2 0 0 mtk_lpm_monitor_thread 0 D LPM-2
root 438 438 2 0 0 mtk_lpm_monitor_thread 0 D LPM-3
root 439 439 2 0 0 mtk_lpm_monitor_thread 0 D LPM-4
root 440 440 2 0 0 mtk_lpm_monitor_thread 0 D LPM-5
root 441 441 2 0 0 mtk_lpm_monitor_thread 0 D LPM-6
root 442 442 2 0 0 mtk_lpm_monitor_thread 0 D LPM-7
Signed-off-by: chenguanyou <[email protected]>
---
include/linux/sched/sysctl.h | 4 ++++
kernel/hung_task.c | 17 +++++++++++++++++
kernel/sysctl.c | 16 ++++++++++++++++
lib/Kconfig.debug | 24 ++++++++++++++++++++++++
4 files changed, 61 insertions(+)
diff --git a/include/linux/sched/sysctl.h b/include/linux/sched/sysctl.h
index 3c31ba88aca5..e8a9a28215bf 100644
--- a/include/linux/sched/sysctl.h
+++ b/include/linux/sched/sysctl.h
@@ -14,11 +14,15 @@ extern unsigned int sysctl_hung_task_all_cpu_backtrace;
#define sysctl_hung_task_all_cpu_backtrace 0
#endif /* CONFIG_SMP */
+#define TASK_COMM_LEN 16
+
extern int sysctl_hung_task_check_count;
extern unsigned int sysctl_hung_task_panic;
extern unsigned long sysctl_hung_task_timeout_secs;
extern unsigned long sysctl_hung_task_check_interval_secs;
extern int sysctl_hung_task_warnings;
+extern unsigned int sysctl_hung_task_filter_kthread;
+extern char sysctl_hung_task_check_comm[TASK_COMM_LEN];
int proc_dohung_task_timeout_secs(struct ctl_table *table, int write,
void *buffer, size_t *lenp, loff_t *ppos);
#else
diff --git a/kernel/hung_task.c b/kernel/hung_task.c
index 396ebaebea3f..e018563d4882 100644
--- a/kernel/hung_task.c
+++ b/kernel/hung_task.c
@@ -23,6 +23,7 @@
#include <linux/sched/sysctl.h>
#include <trace/events/sched.h>
+#include <linux/string.h>
/*
* The number of tasks checked:
@@ -48,6 +49,16 @@ unsigned long __read_mostly sysctl_hung_task_timeout_secs = CONFIG_DEFAULT_HUNG_
*/
unsigned long __read_mostly sysctl_hung_task_check_interval_secs;
+/*
+ * Non zero means no checking kthread
+ */
+unsigned int __read_mostly sysctl_hung_task_filter_kthread = CONFIG_DEFAULT_HUNG_TASK_FILTER_KTHREAD;
+
+/*
+ * Only one
+ */
+char __read_mostly sysctl_hung_task_check_comm[TASK_COMM_LEN] = CONFIG_DEFAULT_HUNG_TASK_CHECK_COMM;
+
int __read_mostly sysctl_hung_task_warnings = 10;
static int __read_mostly did_panic;
@@ -88,6 +99,12 @@ static void check_hung_task(struct task_struct *t, unsigned long timeout)
{
unsigned long switch_count = t->nvcsw + t->nivcsw;
+ if (unlikely(strlen(sysctl_hung_task_check_comm) && !strstr(t->comm, sysctl_hung_task_check_comm)))
+ return;
+
+ if (unlikely(sysctl_hung_task_filter_kthread && t->flags & PF_KTHREAD))
+ return;
+
/*
* Ensure the task is not frozen.
* Also, skip vfork and any other user process that freezer should skip.
diff --git a/kernel/sysctl.c b/kernel/sysctl.c
index 62fbd09b5dc1..1daede87c88d 100644
--- a/kernel/sysctl.c
+++ b/kernel/sysctl.c
@@ -2519,6 +2519,22 @@ static struct ctl_table kern_table[] = {
.proc_handler = proc_dointvec_minmax,
.extra1 = &neg_one,
},
+ {
+ .procname = "hung_task_filter_kthread",
+ .data = &sysctl_hung_task_filter_kthread,
+ .maxlen = sizeof(int),
+ .mode = 0644,
+ .proc_handler = proc_dointvec_minmax,
+ .extra1 = SYSCTL_ZERO,
+ .extra2 = SYSCTL_ONE,
+ },
+ {
+ .procname = "hung_task_check_comm",
+ .data = &sysctl_hung_task_check_comm,
+ .maxlen = TASK_COMM_LEN,
+ .mode = 0644,
+ .proc_handler = proc_dostring,
+ },
#endif
#ifdef CONFIG_RT_MUTEXES
{
diff --git a/lib/Kconfig.debug b/lib/Kconfig.debug
index 2c7f46b366f1..6eab8cf0c37f 100644
--- a/lib/Kconfig.debug
+++ b/lib/Kconfig.debug
@@ -1101,6 +1101,30 @@ config DEFAULT_HUNG_TASK_TIMEOUT
A timeout of 0 disables the check. The default is two minutes.
Keeping the default should be fine in most cases.
+config DEFAULT_HUNG_TASK_FILTER_KTHREAD
+ int "Default filter kthread for hung task"
+ depends on DETECT_HUNG_TASK
+ range 0 1
+ default 0
+ help
+ This option controls filter kthread used to determine when
+ a kernel task has become "state=TASK_UNINTERRUPTIBLE" and should be skip.
+
+ It can be adjusted at runtime via the kernel.hung_task_filter_kthread
+ sysctl or by writing a value to
+ /proc/sys/kernel/hung_task_filter_kthread.
+
+ A filter of 1 disables the check
+
+config DEFAULT_HUNG_TASK_CHECK_COMM
+ string "Default check only one comm"
+ depends on DETECT_HUNG_TASK
+ default ""
+ help
+ It can be adjusted at runtime via the kernel.hung_task_check_comm
+ sysctl or by writing a value to
+ /proc/sys/kernel/hung_task_check_comm.
+
config BOOTPARAM_HUNG_TASK_PANIC
bool "Panic (Reboot) On Hung Tasks"
depends on DETECT_HUNG_TASK
--
2.17.1
On 5/21/21 6:25 AM, chenguanyou wrote:
> Some kernel threads are always in D state, when we enable hung_task,
> it will misjudge, we should skip these to narrow the scope.
>
> exp mtk mobilephone:
> root 420 420 2 0 0 kwdt_thread 0 D wdtk-0
> root 421 421 2 0 0 kwdt_thread 0 D wdtk-1
> root 422 422 2 0 0 kwdt_thread 0 D wdtk-2
> root 423 423 2 0 0 kwdt_thread 0 D wdtk-3
> root 424 424 2 0 0 kwdt_thread 0 D wdtk-4
> root 425 425 2 0 0 kwdt_thread 0 D wdtk-5
> root 426 426 2 0 0 kwdt_thread 0 D wdtk-6
> root 427 427 2 0 0 kwdt_thread 0 D wdtk-7
> root 435 435 2 0 0 mtk_lpm_monitor_thread 0 D LPM-0
> root 436 436 2 0 0 mtk_lpm_monitor_thread 0 D LPM-1
> root 437 437 2 0 0 mtk_lpm_monitor_thread 0 D LPM-2
> root 438 438 2 0 0 mtk_lpm_monitor_thread 0 D LPM-3
> root 439 439 2 0 0 mtk_lpm_monitor_thread 0 D LPM-4
> root 440 440 2 0 0 mtk_lpm_monitor_thread 0 D LPM-5
> root 441 441 2 0 0 mtk_lpm_monitor_thread 0 D LPM-6
> root 442 442 2 0 0 mtk_lpm_monitor_thread 0 D LPM-7
>
> Signed-off-by: chenguanyou <[email protected]>
> ---
> include/linux/sched/sysctl.h | 4 ++++
> kernel/hung_task.c | 17 +++++++++++++++++
> kernel/sysctl.c | 16 ++++++++++++++++
> lib/Kconfig.debug | 24 ++++++++++++++++++++++++
> 4 files changed, 61 insertions(+)
>
> diff --git a/include/linux/sched/sysctl.h b/include/linux/sched/sysctl.h
> index 3c31ba88aca5..e8a9a28215bf 100644
> --- a/include/linux/sched/sysctl.h
> +++ b/include/linux/sched/sysctl.h
> @@ -14,11 +14,15 @@ extern unsigned int sysctl_hung_task_all_cpu_backtrace;
> #define sysctl_hung_task_all_cpu_backtrace 0
> #endif /* CONFIG_SMP */
>
> +#define TASK_COMM_LEN 16
> +
> extern int sysctl_hung_task_check_count;
> extern unsigned int sysctl_hung_task_panic;
> extern unsigned long sysctl_hung_task_timeout_secs;
> extern unsigned long sysctl_hung_task_check_interval_secs;
> extern int sysctl_hung_task_warnings;
> +extern unsigned int sysctl_hung_task_filter_kthread;
> +extern char sysctl_hung_task_check_comm[TASK_COMM_LEN];
> int proc_dohung_task_timeout_secs(struct ctl_table *table, int write,
> void *buffer, size_t *lenp, loff_t *ppos);
> #else
> diff --git a/kernel/hung_task.c b/kernel/hung_task.c
> index 396ebaebea3f..e018563d4882 100644
> --- a/kernel/hung_task.c
> +++ b/kernel/hung_task.c
> @@ -23,6 +23,7 @@
> #include <linux/sched/sysctl.h>
>
> #include <trace/events/sched.h>
> +#include <linux/string.h>
>
> /*
> * The number of tasks checked:
> @@ -48,6 +49,16 @@ unsigned long __read_mostly sysctl_hung_task_timeout_secs = CONFIG_DEFAULT_HUNG_
> */
> unsigned long __read_mostly sysctl_hung_task_check_interval_secs;
>
> +/*
> + * Non zero means no checking kthread
Non-zero
(also mentioned in v1 review, whereas this is v2, I guess, although
it doesn't say that it is v2, but it should say that.)
> + */
> +unsigned int __read_mostly sysctl_hung_task_filter_kthread = CONFIG_DEFAULT_HUNG_TASK_FILTER_KTHREAD;
> +
> +/*
> + * Only one
> + */
> +char __read_mostly sysctl_hung_task_check_comm[TASK_COMM_LEN] = CONFIG_DEFAULT_HUNG_TASK_CHECK_COMM;
> +
> int __read_mostly sysctl_hung_task_warnings = 10;
>
> static int __read_mostly did_panic;
> @@ -88,6 +99,12 @@ static void check_hung_task(struct task_struct *t, unsigned long timeout)
> {
> unsigned long switch_count = t->nvcsw + t->nivcsw;
>
> + if (unlikely(strlen(sysctl_hung_task_check_comm) && !strstr(t->comm, sysctl_hung_task_check_comm)))
> + return;
> +
> + if (unlikely(sysctl_hung_task_filter_kthread && t->flags & PF_KTHREAD))
> + return;
> +
> /*
> * Ensure the task is not frozen.
> * Also, skip vfork and any other user process that freezer should skip.
> diff --git a/kernel/sysctl.c b/kernel/sysctl.c
> index 62fbd09b5dc1..1daede87c88d 100644
> --- a/kernel/sysctl.c
> +++ b/kernel/sysctl.c
> @@ -2519,6 +2519,22 @@ static struct ctl_table kern_table[] = {
> .proc_handler = proc_dointvec_minmax,
> .extra1 = &neg_one,
> },
> + {
> + .procname = "hung_task_filter_kthread",
> + .data = &sysctl_hung_task_filter_kthread,
> + .maxlen = sizeof(int),
> + .mode = 0644,
> + .proc_handler = proc_dointvec_minmax,
> + .extra1 = SYSCTL_ZERO,
> + .extra2 = SYSCTL_ONE,
> + },
> + {
> + .procname = "hung_task_check_comm",
> + .data = &sysctl_hung_task_check_comm,
> + .maxlen = TASK_COMM_LEN,
> + .mode = 0644,
> + .proc_handler = proc_dostring,
> + },
(copy-paste from v1 review:)
These new sysctls should be documented in Documentation/admin-guide/sysctl/kernel.rst.
> #endif
> #ifdef CONFIG_RT_MUTEXES
> {
> diff --git a/lib/Kconfig.debug b/lib/Kconfig.debug
> index 2c7f46b366f1..6eab8cf0c37f 100644
> --- a/lib/Kconfig.debug
> +++ b/lib/Kconfig.debug
> @@ -1101,6 +1101,30 @@ config DEFAULT_HUNG_TASK_TIMEOUT
> A timeout of 0 disables the check. The default is two minutes.
> Keeping the default should be fine in most cases.
>
> +config DEFAULT_HUNG_TASK_FILTER_KTHREAD
> + int "Default filter kthread for hung task"
> + depends on DETECT_HUNG_TASK
> + range 0 1
> + default 0
> + help
> + This option controls filter kthread used to determine when
(again:)
uses
> + a kernel task has become "state=TASK_UNINTERRUPTIBLE" and should be skip.
(again:)
skipped.
> +
> + It can be adjusted at runtime via the kernel.hung_task_filter_kthread
> + sysctl or by writing a value to
> + /proc/sys/kernel/hung_task_filter_kthread.
> +
> + A filter of 1 disables the check
(again:)
check.
> +
> +config DEFAULT_HUNG_TASK_CHECK_COMM
> + string "Default check only one comm"
> + depends on DETECT_HUNG_TASK
> + default ""
> + help
> + It can be adjusted at runtime via the kernel.hung_task_check_comm
> + sysctl or by writing a value to
> + /proc/sys/kernel/hung_task_check_comm.
> +
(again:)
That help text doesn't tell how the Kconfig symbol is used.
> config BOOTPARAM_HUNG_TASK_PANIC
> bool "Panic (Reboot) On Hung Tasks"
> depends on DETECT_HUNG_TASK
>
--
~Randy
On Tue 25-05-21 11:27:16, 陈冠有 wrote:
[...]
> These are MTK's LPM designs.
No idea what those are but it seems like an out of tree code.
> If we have linux-api, hungtask can choose to skip kthread, helpful for us
> to debug user space threads in "state=D" when enable hungtask panic.
No, this approach is wrong. You are trying to workaround an incorrect
out of tree code by creating a user visible API. I have already brought
that up in earlier version of this patch.
Nacked-by: Michal Hocko <[email protected]>
Andrew has already given you a lead on how to fix the said code. Use
TASK_IDLE when waiting for an event. This will both hide that waiting
task from load average accounting and also from the hung task detector.
--
Michal Hocko
SUSE Labs