From 9c44bc03fff44ff04237a7d92e35304a0e50c331 Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Mon, 12 May 2008 21:21:04 +0200 Subject: 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. Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/softlockup.c | 21 +++++++++++++++++++++ kernel/sysctl.c | 11 +++++++++++ 2 files changed, 32 insertions(+) (limited to 'kernel') diff --git a/kernel/softlockup.c b/kernel/softlockup.c index 01b6522fd92b..78e0ad21cb0c 100644 --- a/kernel/softlockup.c +++ b/kernel/softlockup.c @@ -27,6 +27,21 @@ static DEFINE_PER_CPU(struct task_struct *, watchdog_task); static int __read_mostly did_panic; unsigned long __read_mostly softlockup_thresh = 60; +/* + * Should we panic (and reboot, if panic_timeout= is set) when a + * soft-lockup occurs: + */ +unsigned int __read_mostly softlockup_panic = + CONFIG_BOOTPARAM_SOFTLOCKUP_PANIC_VALUE; + +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 softlock_panic(struct notifier_block *this, unsigned long event, void *ptr) { @@ -120,6 +135,9 @@ void softlockup_tick(void) else dump_stack(); spin_unlock(&print_lock); + + if (softlockup_panic) + panic("softlockup: hung tasks"); } /* @@ -172,6 +190,9 @@ static void check_hung_task(struct task_struct *t, unsigned long now) t->last_switch_timestamp = now; touch_nmi_watchdog(); + + if (softlockup_panic) + panic("softlockup: blocked tasks"); } /* diff --git a/kernel/sysctl.c b/kernel/sysctl.c index 29116652dca8..2d3b388c402d 100644 --- a/kernel/sysctl.c +++ b/kernel/sysctl.c @@ -727,6 +727,17 @@ static struct ctl_table kern_table[] = { }, #endif #ifdef CONFIG_DETECT_SOFTLOCKUP + { + .ctl_name = CTL_UNNUMBERED, + .procname = "softlockup_panic", + .data = &softlockup_panic, + .maxlen = sizeof(int), + .mode = 0644, + .proc_handler = &proc_doulongvec_minmax, + .strategy = &sysctl_intvec, + .extra1 = &zero, + .extra2 = &one, + }, { .ctl_name = CTL_UNNUMBERED, .procname = "softlockup_thresh", -- cgit v1.2.2 From 9383d9679056e6cc4e7ff70f31da945a268238f4 Mon Sep 17 00:00:00 2001 From: Dimitri Sivanich Date: Mon, 12 May 2008 21:21:14 +0200 Subject: softlockup: fix softlockup_thresh unaligned access and disable detection at runtime Fix unaligned access errors when setting softlockup_thresh on 64 bit platforms. Allow softlockup detection to be disabled by setting softlockup_thresh <= 0. Detect that boot time softlockup detection has been disabled earlier in softlockup_tick. Signed-off-by: Dimitri Sivanich Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/softlockup.c | 12 ++++++++++-- kernel/sysctl.c | 9 +++++---- 2 files changed, 15 insertions(+), 6 deletions(-) (limited to 'kernel') diff --git a/kernel/softlockup.c b/kernel/softlockup.c index 78e0ad21cb0c..a3a0b239b7f7 100644 --- a/kernel/softlockup.c +++ b/kernel/softlockup.c @@ -25,7 +25,7 @@ static DEFINE_PER_CPU(unsigned long, print_timestamp); static DEFINE_PER_CPU(struct task_struct *, watchdog_task); static int __read_mostly did_panic; -unsigned long __read_mostly softlockup_thresh = 60; +int __read_mostly softlockup_thresh = 60; /* * Should we panic (and reboot, if panic_timeout= is set) when a @@ -94,6 +94,14 @@ void softlockup_tick(void) struct pt_regs *regs = get_irq_regs(); unsigned long now; + /* Is detection switched off? */ + if (!per_cpu(watchdog_task, this_cpu) || softlockup_thresh <= 0) { + /* Be sure we don't false trigger if switched back on */ + if (touch_timestamp) + per_cpu(touch_timestamp, this_cpu) = 0; + return; + } + if (touch_timestamp == 0) { touch_softlockup_watchdog(); return; @@ -104,7 +112,7 @@ void softlockup_tick(void) /* report at most once a second */ if ((print_timestamp >= touch_timestamp && print_timestamp < (touch_timestamp + 1)) || - did_panic || !per_cpu(watchdog_task, this_cpu)) { + did_panic) { return; } diff --git a/kernel/sysctl.c b/kernel/sysctl.c index 2d3b388c402d..31c19a79738d 100644 --- a/kernel/sysctl.c +++ b/kernel/sysctl.c @@ -84,12 +84,13 @@ extern int latencytop_enabled; extern int sysctl_nr_open_min, sysctl_nr_open_max; /* Constants used for minimum and maximum */ -#if defined(CONFIG_DETECT_SOFTLOCKUP) || defined(CONFIG_HIGHMEM) +#ifdef CONFIG_HIGHMEM static int one = 1; #endif #ifdef CONFIG_DETECT_SOFTLOCKUP static int sixty = 60; +static int neg_one = -1; #endif #ifdef CONFIG_MMU @@ -742,11 +743,11 @@ static struct ctl_table kern_table[] = { .ctl_name = CTL_UNNUMBERED, .procname = "softlockup_thresh", .data = &softlockup_thresh, - .maxlen = sizeof(unsigned long), + .maxlen = sizeof(int), .mode = 0644, - .proc_handler = &proc_doulongvec_minmax, + .proc_handler = &proc_dointvec_minmax, .strategy = &sysctl_intvec, - .extra1 = &one, + .extra1 = &neg_one, .extra2 = &sixty, }, { -- cgit v1.2.2 From 1c4cd6dd1d0fd3057bb6b8c87460049497889d1b Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Mon, 12 May 2008 21:21:14 +0200 Subject: softlockup: fix softlockup_thresh fix Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/sysctl.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/sysctl.c b/kernel/sysctl.c index 31c19a79738d..a829dc8d7a9e 100644 --- a/kernel/sysctl.c +++ b/kernel/sysctl.c @@ -84,7 +84,7 @@ extern int latencytop_enabled; extern int sysctl_nr_open_min, sysctl_nr_open_max; /* Constants used for minimum and maximum */ -#ifdef CONFIG_HIGHMEM +#if defined(CONFIG_HIGHMEM) || defined(CONFIG_DETECT_SOFTLOCKUP) static int one = 1; #endif -- cgit v1.2.2 From 02ff375590ac4140d88afc76505df1ad45c6af59 Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Mon, 12 May 2008 15:43:53 +0200 Subject: softlockup: fix false positives on nohz if CPU is 100% idle for more than 60 seconds Fix (probably theoretical only) rq->clock update bug: in tick_nohz_update_jiffies() [which is called on all irq entry on all cpus where the irq entry hits an idle cpu] we call touch_softlockup_watchdog() before we update jiffies. That works fine most of the time when idle timeouts are within 60 seconds. But when an idle timeout is beyond 60 seconds, jiffies is updated with a jump of more than 60 seconds, which causes a jump in cpu-clock of more than 60 seconds, triggering a false positive. Reported-by: David Miller Signed-off-by: Ingo Molnar --- kernel/time/tick-sched.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'kernel') diff --git a/kernel/time/tick-sched.c b/kernel/time/tick-sched.c index b854a895591e..28abad66fc8e 100644 --- a/kernel/time/tick-sched.c +++ b/kernel/time/tick-sched.c @@ -133,8 +133,6 @@ void tick_nohz_update_jiffies(void) if (!ts->tick_stopped) return; - touch_softlockup_watchdog(); - cpu_clear(cpu, nohz_cpu_mask); now = ktime_get(); ts->idle_waketime = now; @@ -142,6 +140,8 @@ void tick_nohz_update_jiffies(void) local_irq_save(flags); tick_do_update_jiffies64(now); local_irq_restore(flags); + + touch_softlockup_watchdog(); } void tick_nohz_stop_idle(int cpu) -- cgit v1.2.2 From 8c2238eaaf0f774ca0f8d9daad7a616429bbb7f1 Mon Sep 17 00:00:00 2001 From: Jason Wessel Date: Tue, 27 May 2008 12:23:29 -0500 Subject: softlockup: fix NMI hangs due to lock race - 2.6.26-rc regression The touch_nmi_watchdog() routine on x86 ultimately calls touch_softlockup_watchdog(). The problem is that to touch the softlockup watchdog, the cpu_clock code has to be called which could involve multiple cpu locks and can lead to a hard hang if one of the locks is held by a processor that is not going to return anytime soon (such as could be the case with kgdb or perhaps even with some other kind of exception). This patch causes the public version of the touch_softlockup_watchdog() to defer the cpu clock access to a later point. The test case for this problem is to use the following kernel config options: CONFIG_KGDB_TESTS=y CONFIG_KGDB_TESTS_ON_BOOT=y CONFIG_KGDB_TESTS_BOOT_STRING="V1F100I100000" It should be noted that kgdb test suite and these options were not available until 2.6.26-rc2, so it was necessary to patch the kgdb test suite during the bisection. I would consider this patch a regression fix because the problem first appeared in commit 27ec4407790d075c325e1f4da0a19c56953cce23 when some logic was added to try to periodically sync the clocks. It was possible to work around this particular problem by simply not performing the sync anytime the system was in a critical context. This was ok until commit 3e51f33fcc7f55e6df25d15b55ed10c8b4da84cd, which added config option CONFIG_HAVE_UNSTABLE_SCHED_CLOCK and some multi-cpu locks to sync the clocks. It became clear that accessing this code from an nmi was the source of the lockups. Avoiding the access to the low level clock code from an code inside the NMI processing also fixed the problem with the 27ec44... commit. Signed-off-by: Jason Wessel Signed-off-by: Ingo Molnar --- kernel/softlockup.c | 15 ++++++++++----- 1 file changed, 10 insertions(+), 5 deletions(-) (limited to 'kernel') diff --git a/kernel/softlockup.c b/kernel/softlockup.c index a3a0b239b7f7..6b682d86bddf 100644 --- a/kernel/softlockup.c +++ b/kernel/softlockup.c @@ -64,12 +64,17 @@ static unsigned long get_timestamp(int this_cpu) return cpu_clock(this_cpu) >> 30LL; /* 2^30 ~= 10^9 */ } -void touch_softlockup_watchdog(void) +static void __touch_softlockup_watchdog(void) { int this_cpu = raw_smp_processor_id(); __raw_get_cpu_var(touch_timestamp) = get_timestamp(this_cpu); } + +void touch_softlockup_watchdog(void) +{ + __raw_get_cpu_var(touch_timestamp) = 0; +} EXPORT_SYMBOL(touch_softlockup_watchdog); void touch_all_softlockup_watchdogs(void) @@ -103,7 +108,7 @@ void softlockup_tick(void) } if (touch_timestamp == 0) { - touch_softlockup_watchdog(); + __touch_softlockup_watchdog(); return; } @@ -118,7 +123,7 @@ void softlockup_tick(void) /* do not print during early bootup: */ if (unlikely(system_state != SYSTEM_RUNNING)) { - touch_softlockup_watchdog(); + __touch_softlockup_watchdog(); return; } @@ -243,7 +248,7 @@ static int watchdog(void *__bind_cpu) sched_setscheduler(current, SCHED_FIFO, ¶m); /* initialize timestamp */ - touch_softlockup_watchdog(); + __touch_softlockup_watchdog(); set_current_state(TASK_INTERRUPTIBLE); /* @@ -252,7 +257,7 @@ static int watchdog(void *__bind_cpu) * debug-printout triggers in softlockup_tick(). */ while (!kthread_should_stop()) { - touch_softlockup_watchdog(); + __touch_softlockup_watchdog(); schedule(); if (kthread_should_stop()) -- cgit v1.2.2 From 688c91755dc3d3c03d8c67c1df13c02be258768e Mon Sep 17 00:00:00 2001 From: Arjan van de Ven Date: Mon, 16 Jun 2008 15:51:08 -0700 Subject: softlockup: print a module list on being stuck Most places in the kernel that go BUG: print a module list (which is very useful for doing statistics and finding patterns), however the softlockup detector does not do this yet. This patch adds the one line change to fix this gap. Signed-off-by: Arjan van de Ven Signed-off-by: Ingo Molnar --- kernel/softlockup.c | 1 + 1 file changed, 1 insertion(+) (limited to 'kernel') diff --git a/kernel/softlockup.c b/kernel/softlockup.c index 6b682d86bddf..f2bf5decb108 100644 --- a/kernel/softlockup.c +++ b/kernel/softlockup.c @@ -143,6 +143,7 @@ void softlockup_tick(void) printk(KERN_ERR "BUG: soft lockup - CPU#%d stuck for %lus! [%s:%d]\n", this_cpu, now - touch_timestamp, current->comm, task_pid_nr(current)); + print_modules(); if (regs) show_regs(regs); else -- cgit v1.2.2 From 8d5be7f4e8515af461cbc8f07687ccc81507d508 Mon Sep 17 00:00:00 2001 From: Vegard Nossum Date: Wed, 25 Jun 2008 08:50:10 +0200 Subject: softlockup: show irqtrace This patch adds some information about when interrupts were last enabled and disabled to the output of the softlockup detector. Signed-off-by: Vegard Nossum Cc: Peter Zijlstra Cc: Johannes Weiner Cc: Arjan van de Ven Signed-off-by: Ingo Molnar --- kernel/softlockup.c | 2 ++ 1 file changed, 2 insertions(+) (limited to 'kernel') diff --git a/kernel/softlockup.c b/kernel/softlockup.c index f2bf5decb108..97977ecc3171 100644 --- a/kernel/softlockup.c +++ b/kernel/softlockup.c @@ -13,6 +13,7 @@ #include #include #include +#include #include #include @@ -144,6 +145,7 @@ void softlockup_tick(void) this_cpu, now - touch_timestamp, current->comm, task_pid_nr(current)); print_modules(); + print_irqtrace_events(current); if (regs) show_regs(regs); else -- cgit v1.2.2 From dd7a1e5615b1719c0fdffee1ea5a7820ac8141a6 Mon Sep 17 00:00:00 2001 From: Johannes Weiner Date: Fri, 27 Jun 2008 15:07:21 +0200 Subject: softlockup: fix watchdog task wakeup frequency Updating the timestamp more often is pointless as we print the warnings only if we exceed the threshold. And the check for hung tasks relies on the last timestamp, so it will keep working correctly, too. Signed-off-by: Johannes Weiner Cc: Peter Zijlstra Signed-off-by: Ingo Molnar --- kernel/softlockup.c | 7 +++++-- 1 file changed, 5 insertions(+), 2 deletions(-) (limited to 'kernel') diff --git a/kernel/softlockup.c b/kernel/softlockup.c index 97977ecc3171..d53ab702a7db 100644 --- a/kernel/softlockup.c +++ b/kernel/softlockup.c @@ -130,8 +130,11 @@ void softlockup_tick(void) now = get_timestamp(this_cpu); - /* Wake up the high-prio watchdog task every second: */ - if (now > (touch_timestamp + 1)) + /* + * Wake up the high-prio watchdog task twice per + * threshold timespan. + */ + if (now > touch_timestamp + softlockup_thresh/2) wake_up_process(per_cpu(watchdog_task, this_cpu)); /* Warn about unreasonable delays: */ -- cgit v1.2.2 From 3e2f69fdd1b00166e7d589bce56b2d36a9e74374 Mon Sep 17 00:00:00 2001 From: Johannes Weiner Date: Tue, 1 Jul 2008 09:12:04 +0200 Subject: softlockup: fix watchdog task wakeup frequency The print_timestamp can never be bigger than the touch_timestamp, at maximum it can be equal. And if it is, the second check for touch_timestamp + 1 bigger print_timestamp is always true, too. The check for equality is sufficient as we proceed in one-second-steps and are at least one second away from the last print-out if we have another timestamp. Signed-off-by: Johannes Weiner Cc: Peter Zijlstra Signed-off-by: Ingo Molnar --- kernel/softlockup.c | 5 +---- 1 file changed, 1 insertion(+), 4 deletions(-) (limited to 'kernel') diff --git a/kernel/softlockup.c b/kernel/softlockup.c index d53ab702a7db..7bd8d1aadd5d 100644 --- a/kernel/softlockup.c +++ b/kernel/softlockup.c @@ -116,11 +116,8 @@ void softlockup_tick(void) print_timestamp = per_cpu(print_timestamp, this_cpu); /* report at most once a second */ - if ((print_timestamp >= touch_timestamp && - print_timestamp < (touch_timestamp + 1)) || - did_panic) { + if (print_timestamp == touch_timestamp || did_panic) return; - } /* do not print during early bootup: */ if (unlikely(system_state != SYSTEM_RUNNING)) { -- cgit v1.2.2 From 4dca10a96041f78bed11ce9e4a5cfde813ec4ccb Mon Sep 17 00:00:00 2001 From: Hiroshi Shimamoto Date: Mon, 7 Jul 2008 18:37:04 -0700 Subject: softlockup: fix invalid proc_handler for softlockup_panic The type of softlockup_panic is int, but the proc_handler is proc_doulongvec_minmax(). This handler is for unsigned long. This handler should be proc_dointvec_minmax(). Signed-off-by: Hiroshi Shimamoto Signed-off-by: Ingo Molnar --- kernel/sysctl.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/sysctl.c b/kernel/sysctl.c index 3e56d2f91414..ab59ac008caf 100644 --- a/kernel/sysctl.c +++ b/kernel/sysctl.c @@ -741,7 +741,7 @@ static struct ctl_table kern_table[] = { .data = &softlockup_panic, .maxlen = sizeof(int), .mode = 0644, - .proc_handler = &proc_doulongvec_minmax, + .proc_handler = &proc_dointvec_minmax, .strategy = &sysctl_intvec, .extra1 = &zero, .extra2 = &one, -- cgit v1.2.2