diff options
| -rw-r--r-- | arch/x86/kernel/apic/hw_nmi.c | 4 | ||||
| -rw-r--r-- | include/linux/nmi.h | 2 | ||||
| -rw-r--r-- | kernel/watchdog.c | 19 |
3 files changed, 18 insertions, 7 deletions
diff --git a/arch/x86/kernel/apic/hw_nmi.c b/arch/x86/kernel/apic/hw_nmi.c index 5260fe91bcb6..d5e57db0f7be 100644 --- a/arch/x86/kernel/apic/hw_nmi.c +++ b/arch/x86/kernel/apic/hw_nmi.c | |||
| @@ -19,9 +19,9 @@ | |||
| 19 | #include <linux/delay.h> | 19 | #include <linux/delay.h> |
| 20 | 20 | ||
| 21 | #ifdef CONFIG_HARDLOCKUP_DETECTOR | 21 | #ifdef CONFIG_HARDLOCKUP_DETECTOR |
| 22 | u64 hw_nmi_get_sample_period(void) | 22 | u64 hw_nmi_get_sample_period(int watchdog_thresh) |
| 23 | { | 23 | { |
| 24 | return (u64)(cpu_khz) * 1000 * 60; | 24 | return (u64)(cpu_khz) * 1000 * watchdog_thresh; |
| 25 | } | 25 | } |
| 26 | #endif | 26 | #endif |
| 27 | 27 | ||
diff --git a/include/linux/nmi.h b/include/linux/nmi.h index 5317b8b2198f..2d304efc89df 100644 --- a/include/linux/nmi.h +++ b/include/linux/nmi.h | |||
| @@ -45,7 +45,7 @@ static inline bool trigger_all_cpu_backtrace(void) | |||
| 45 | 45 | ||
| 46 | #ifdef CONFIG_LOCKUP_DETECTOR | 46 | #ifdef CONFIG_LOCKUP_DETECTOR |
| 47 | int hw_nmi_is_cpu_stuck(struct pt_regs *); | 47 | int hw_nmi_is_cpu_stuck(struct pt_regs *); |
| 48 | u64 hw_nmi_get_sample_period(void); | 48 | u64 hw_nmi_get_sample_period(int watchdog_thresh); |
| 49 | extern int watchdog_enabled; | 49 | extern int watchdog_enabled; |
| 50 | extern int watchdog_thresh; | 50 | extern int watchdog_thresh; |
| 51 | struct ctl_table; | 51 | struct ctl_table; |
diff --git a/kernel/watchdog.c b/kernel/watchdog.c index 60301916f62e..6e63097fa73a 100644 --- a/kernel/watchdog.c +++ b/kernel/watchdog.c | |||
| @@ -28,7 +28,7 @@ | |||
| 28 | #include <linux/perf_event.h> | 28 | #include <linux/perf_event.h> |
| 29 | 29 | ||
| 30 | int watchdog_enabled = 1; | 30 | int watchdog_enabled = 1; |
| 31 | int __read_mostly watchdog_thresh = 60; | 31 | int __read_mostly watchdog_thresh = 10; |
| 32 | 32 | ||
| 33 | static DEFINE_PER_CPU(unsigned long, watchdog_touch_ts); | 33 | static DEFINE_PER_CPU(unsigned long, watchdog_touch_ts); |
| 34 | static DEFINE_PER_CPU(struct task_struct *, softlockup_watchdog); | 34 | static DEFINE_PER_CPU(struct task_struct *, softlockup_watchdog); |
| @@ -91,6 +91,17 @@ static int __init nosoftlockup_setup(char *str) | |||
| 91 | __setup("nosoftlockup", nosoftlockup_setup); | 91 | __setup("nosoftlockup", nosoftlockup_setup); |
| 92 | /* */ | 92 | /* */ |
| 93 | 93 | ||
| 94 | /* | ||
| 95 | * Hard-lockup warnings should be triggered after just a few seconds. Soft- | ||
| 96 | * lockups can have false positives under extreme conditions. So we generally | ||
| 97 | * want a higher threshold for soft lockups than for hard lockups. So we couple | ||
| 98 | * the thresholds with a factor: we make the soft threshold twice the amount of | ||
| 99 | * time the hard threshold is. | ||
| 100 | */ | ||
| 101 | static int get_softlockup_thresh() | ||
| 102 | { | ||
| 103 | return watchdog_thresh * 2; | ||
| 104 | } | ||
| 94 | 105 | ||
| 95 | /* | 106 | /* |
| 96 | * Returns seconds, approximately. We don't need nanosecond | 107 | * Returns seconds, approximately. We don't need nanosecond |
| @@ -110,7 +121,7 @@ static unsigned long get_sample_period(void) | |||
| 110 | * increment before the hardlockup detector generates | 121 | * increment before the hardlockup detector generates |
| 111 | * a warning | 122 | * a warning |
| 112 | */ | 123 | */ |
| 113 | return watchdog_thresh * (NSEC_PER_SEC / 5); | 124 | return get_softlockup_thresh() * (NSEC_PER_SEC / 5); |
| 114 | } | 125 | } |
| 115 | 126 | ||
| 116 | /* Commands for resetting the watchdog */ | 127 | /* Commands for resetting the watchdog */ |
| @@ -182,7 +193,7 @@ static int is_softlockup(unsigned long touch_ts) | |||
| 182 | unsigned long now = get_timestamp(smp_processor_id()); | 193 | unsigned long now = get_timestamp(smp_processor_id()); |
| 183 | 194 | ||
| 184 | /* Warn about unreasonable delays: */ | 195 | /* Warn about unreasonable delays: */ |
| 185 | if (time_after(now, touch_ts + watchdog_thresh)) | 196 | if (time_after(now, touch_ts + get_softlockup_thresh())) |
| 186 | return now - touch_ts; | 197 | return now - touch_ts; |
| 187 | 198 | ||
| 188 | return 0; | 199 | return 0; |
| @@ -359,7 +370,7 @@ static int watchdog_nmi_enable(int cpu) | |||
| 359 | 370 | ||
| 360 | /* Try to register using hardware perf events */ | 371 | /* Try to register using hardware perf events */ |
| 361 | wd_attr = &wd_hw_attr; | 372 | wd_attr = &wd_hw_attr; |
| 362 | wd_attr->sample_period = hw_nmi_get_sample_period(); | 373 | wd_attr->sample_period = hw_nmi_get_sample_period(watchdog_thresh); |
| 363 | event = perf_event_create_kernel_counter(wd_attr, cpu, NULL, watchdog_overflow_callback); | 374 | event = perf_event_create_kernel_counter(wd_attr, cpu, NULL, watchdog_overflow_callback); |
| 364 | if (!IS_ERR(event)) { | 375 | if (!IS_ERR(event)) { |
| 365 | printk(KERN_INFO "NMI watchdog enabled, takes one hw-pmu counter.\n"); | 376 | printk(KERN_INFO "NMI watchdog enabled, takes one hw-pmu counter.\n"); |
