watchdog/softlockup: use printk_cpu_sync_get_irqsave() to serialize reporting
authorDouglas Anderson <dianders@chromium.org>
Wed, 20 Dec 2023 21:15:35 +0000 (13:15 -0800)
committerAndrew Morton <akpm@linux-foundation.org>
Fri, 29 Dec 2023 20:22:30 +0000 (12:22 -0800)
Instead of introducing a spinlock, use printk_cpu_sync_get_irqsave() and
printk_cpu_sync_put_irqrestore() to serialize softlockup reporting.  Alone
this doesn't have any real advantage over the spinlock, but this will
allow us to use the same function in a future change to also serialize
hardlockup crawls.

NOTE: for the most part this serialization is important because we often
end up in the show_regs() path and that has no built-in serialization if
there are multiple callers at once.  However, even in the case where we
end up in the dump_stack() path this still has some advantages because the
stack will be guaranteed to be together in the logs with the lockup
message with no interleaving.

NOTE: the fact that printk_cpu_sync_get_irqsave() is allowed to be called
multiple times on the same CPU is important here.  Specifically we hold
the "lock" while calling dump_stack() which also gets the same "lock".
This is explicitly documented to be OK and means we don't need to
introduce a variant of dump_stack() that doesn't grab the lock.

Link: https://lkml.kernel.org/r/20231220131534.2.Ia5906525d440d8e8383cde31b7c61c2aadc8f907@changeid
Signed-off-by: Douglas Anderson <dianders@chromium.org>
Reviewed-by: John Ogness <john.ogness@linutronix.de>
Reviewed-by: Li Zhe <lizhe.67@bytedance.com>
Cc: Lecopzer Chen <lecopzer.chen@mediatek.com>
Cc: Petr Mladek <pmladek@suse.com>
Cc: Pingfan Liu <kernelfans@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
kernel/watchdog.c

index b4fd2f12137f000f600b91200f9366fe6f33d20a..526041a1100ace3624e7bced17b096cfb0243ebd 100644 (file)
@@ -454,7 +454,7 @@ static enum hrtimer_restart watchdog_timer_fn(struct hrtimer *hrtimer)
        struct pt_regs *regs = get_irq_regs();
        int duration;
        int softlockup_all_cpu_backtrace = sysctl_softlockup_all_cpu_backtrace;
-       static DEFINE_SPINLOCK(watchdog_output_lock);
+       unsigned long flags;
 
        if (!watchdog_enabled)
                return HRTIMER_NORESTART;
@@ -521,7 +521,7 @@ static enum hrtimer_restart watchdog_timer_fn(struct hrtimer *hrtimer)
                /* Start period for the next softlockup warning. */
                update_report_ts();
 
-               spin_lock(&watchdog_output_lock);
+               printk_cpu_sync_get_irqsave(flags);
                pr_emerg("BUG: soft lockup - CPU#%d stuck for %us! [%s:%d]\n",
                        smp_processor_id(), duration,
                        current->comm, task_pid_nr(current));
@@ -531,7 +531,7 @@ static enum hrtimer_restart watchdog_timer_fn(struct hrtimer *hrtimer)
                        show_regs(regs);
                else
                        dump_stack();
-               spin_unlock(&watchdog_output_lock);
+               printk_cpu_sync_put_irqrestore(flags);
 
                if (softlockup_all_cpu_backtrace) {
                        trigger_allbutcpu_cpu_backtrace(smp_processor_id());