softlockup: allow panic on lockup

allow users to configure the softlockup detector to generate a panic instead of a warning message. high-availability systems might opt for this strict method (combined with panic_timeout= boot option/sysctl), instead of generating softlockup warnings ad infinitum. also, automated tests work better if the system reboots reliably (into a safe kernel) in case of a lockup. The full spectrum of configurability is supported: boot option, sysctl option and Kconfig option. it's default-disabled.

Ingo Molnar committed 18 years ago

Commit 9c44bc01 parent 75d3bce5 files · +62 −2Browse files

Unified · Split

5 files

Documentation/kernel-parameters.txt +3 −0
include/linux/sched.h +2 −1
kernel/softlockup.c +21 −0
kernel/sysctl.c +11 −0
lib/Kconfig.debug +25 −1

Documentation/kernel-parameters.txt+3−0

@@ -1971,6 +1971,9 @@ and is between 256 and 4096 characters. It is defined in the file · 1970 unmodified lines
1971		1971
1972	snd-ymfpci= [HW,ALSA]	1972	snd-ymfpci= [HW,ALSA]
1973		1973
		1974	softlockup_panic=
		1975	[KNL] Should the soft-lockup detector generate panics.
		1976
1974	sonypi.*= [HW] Sony Programmable I/O Control Device driver	1977	sonypi.*= [HW] Sony Programmable I/O Control Device driver
1975	See Documentation/sonypi.txt	1978	See Documentation/sonypi.txt
1976		1979

include/linux/sched.h+2−1

 extern void spawn_softlockup_task(void);
 extern void touch_softlockup_watchdog(void);
 extern void touch_all_softlockup_watchdogs(void);
-extern unsigned long  softlockup_thresh;
+extern unsigned int  softlockup_panic;
+extern unsigned long softlockup_thresh;
 extern unsigned long sysctl_hung_task_check_count;
 extern unsigned long sysctl_hung_task_timeout_secs;
 extern unsigned long sysctl_hung_task_warnings;

kernel/softlockup.c+21−0

@@ -27,6 +27,21 @@ static DEFINE_PER_CPU(struct task_struct *, watchdog_task); · 26 unmodified lines
27	static int __read_mostly did_panic;	27	static int __read_mostly did_panic;
28	unsigned long __read_mostly softlockup_thresh = 60;	28	unsigned long __read_mostly softlockup_thresh = 60;
29		29
		30	/*
		31	* Should we panic (and reboot, if panic_timeout= is set) when a
		32	* soft-lockup occurs:
		33	*/
		34	unsigned int __read_mostly softlockup_panic =
		35	CONFIG_BOOTPARAM_SOFTLOCKUP_PANIC_VALUE;
		36
		37	static int __init softlockup_panic_setup(char *str)
		38	{
		39	softlockup_panic = simple_strtoul(str, NULL, 0);
		40
		41	return 1;
		42	}
		43	__setup("softlockup_panic=", softlockup_panic_setup);
		44
30	static int	45	static int
31	softlock_panic(struct notifier_block this, unsigned long event, void ptr)	46	softlock_panic(struct notifier_block this, unsigned long event, void ptr)
32	{	47	{
@@ -120,6 +135,9 @@ void softlockup_tick(void) · 87 unmodified lines
120	else	135	else
121	dump_stack();	136	dump_stack();
122	spin_unlock(&print_lock);	137	spin_unlock(&print_lock);
		138
		139	if (softlockup_panic)
		140	panic("softlockup: hung tasks");
123	}	141	}
124		142
125	/*	143	/*
@@ -172,6 +190,9 @@ static void check_hung_task(struct task_struct *t, unsigned long now) · 46 unmodified lines
172		190
173	t->last_switch_timestamp = now;	191	t->last_switch_timestamp = now;
174	touch_nmi_watchdog();	192	touch_nmi_watchdog();
		193
		194	if (softlockup_panic)
		195	panic("softlockup: blocked tasks");
175	}	196	}
176		197
177	/*	198	/*

kernel/sysctl.c+11−0

@@ -727,6 +727,17 @@ static struct ctl_table kern_table[] = { · 726 unmodified lines
727	},	727	},
728	#endif	728	#endif
729	#ifdef CONFIG_DETECT_SOFTLOCKUP	729	#ifdef CONFIG_DETECT_SOFTLOCKUP
		730	{
		731	.ctl_name = CTL_UNNUMBERED,
		732	.procname = "softlockup_panic",
		733	.data = &softlockup_panic,
		734	.maxlen = sizeof(int),
		735	.mode = 0644,
		736	.proc_handler = &proc_doulongvec_minmax,
		737	.strategy = &sysctl_intvec,
		738	.extra1 = &zero,
		739	.extra2 = &one,
		740	},
730	{	741	{
731	.ctl_name = CTL_UNNUMBERED,	742	.ctl_name = CTL_UNNUMBERED,
732	.procname = "softlockup_thresh",	743	.procname = "softlockup_thresh",

lib/Kconfig.debug+25−1

@@ -147,7 +147,7 @@ config DETECT_SOFTLOCKUP · 146 unmodified lines
147	help	147	help
148	Say Y here to enable the kernel to detect "soft lockups",	148	Say Y here to enable the kernel to detect "soft lockups",
149	which are bugs that cause the kernel to loop in kernel	149	which are bugs that cause the kernel to loop in kernel
150	mode for more than 10 seconds, without giving other tasks a	150	mode for more than 60 seconds, without giving other tasks a
151	chance to run.	151	chance to run.
152		152
153	When a soft-lockup is detected, the kernel will print the	153	When a soft-lockup is detected, the kernel will print the
@@ -159,6 +159,30 @@ config DETECT_SOFTLOCKUP · 5 unmodified lines
159	can be detected via the NMI-watchdog, on platforms that	159	can be detected via the NMI-watchdog, on platforms that
160	support it.)	160	support it.)
161		161
		162	config BOOTPARAM_SOFTLOCKUP_PANIC
		163	bool "Panic (Reboot) On Soft Lockups"
		164	depends on DETECT_SOFTLOCKUP
		165	help
		166	Say Y here to enable the kernel to panic on "soft lockups",
		167	which are bugs that cause the kernel to loop in kernel
		168	mode for more than 60 seconds, without giving other tasks a
		169	chance to run.
		170
		171	The panic can be used in combination with panic_timeout,
		172	to cause the system to reboot automatically after a
		173	lockup has been detected. This feature is useful for
		174	high-availability systems that have uptime guarantees and
		175	where a lockup must be resolved ASAP.
		176
		177	Say N if unsure.
		178
		179	config BOOTPARAM_SOFTLOCKUP_PANIC_VALUE
		180	int
		181	depends on DETECT_SOFTLOCKUP
		182	range 0 1
		183	default 0 if !BOOTPARAM_SOFTLOCKUP_PANIC
		184	default 1 if BOOTPARAM_SOFTLOCKUP_PANIC
		185
162	config SCHED_DEBUG	186	config SCHED_DEBUG
163	bool "Collect scheduler debugging info"	187	bool "Collect scheduler debugging info"
164	depends on DEBUG_KERNEL && PROC_FS	188	depends on DEBUG_KERNEL && PROC_FS

softlockup: allow panic on lockup

Keyboard shortcuts

General

Issues and pull requests

Issues, pull requests and Inbox

Repository code