softlockup 解决思路

 在修改oss驱动的时候，出现了soft lockup detected 的错误，查看一下内核源码，基本确定了导致问题的原因。内核里有一个watch内核线程，watchdog()函数是线程函数，这个函数没一秒被调用一次，在函数中更新一个时间戳。softlockup_tick()函数在每一个tick都被调用一次，在这个函数中会检查是否该唤醒watchdog内核线程，并检查是否发生soft lockup。如果有10秒或更长时间watchdog线程没有被调用，就认为发生soft lockup。当一个低优先级的中断到来的时候，会屏蔽时钟中断。cpu在处理中断时是关中断运行的，即使是时钟中断也被屏蔽。所以watchdog（）函数不能运行，时间戳自然就没有被更新，时间太长的话，就soft lockup了。有可能是某个驱动在中断处理中处理时间太久，比如循环，spinlock使用不当，都会造成soft lockup的发生。下面给出详细代码，免的在内核找来找去.
kernel/softlockup.c
/*
 * Detect Soft Lockups
 *
 * started by Ingo Molnar, Copyright (C) 2005, 2006 Red Hat, Inc.
 *
 * this code detects soft lockups: incidents in where on a CPU
 * the kernel does not reschedule for 10 seconds or more.
 */
#include
#include
#include
#include
#include
#include
#include

static DEFINE_SPINLOCK(print_lock);

static DEFINE_PER_CPU(unsigned long, touch_timestamp);
static DEFINE_PER_CPU(unsigned long, print_timestamp);
static DEFINE_PER_CPU(struct task_struct *, watchdog_task);

static int did_panic = 0;

static int
softlock_panic(struct notifier_block *this, unsigned long event, void *ptr)
{
    did_panic = 1;

    return NOTIFY_DONE;
}

static struct notifier_block panic_block = {
    .notifier_call = softlock_panic,
};

void touch_softlockup_watchdog(void)
{
    __raw_get_cpu_var(touch_timestamp) = jiffies;
}
EXPORT_SYMBOL(touch_softlockup_watchdog);

/*
 * This callback runs from the timer interrupt, and checks
 * whether the watchdog thread has hung or not:
 */
void softlockup_tick(void)
{
    int this_cpu = smp_processor_id();
    unsigned long touch_timestamp = per_cpu(touch_timestamp, this_cpu);

    /* prevent double reports: */
    if (per_cpu(print_timestamp, this_cpu) == touch_timestamp ||
        did_panic ||
            !per_cpu(watchdog_task, this_cpu))
        return;

    /* do not print during early bootup: */
    if (unlikely(system_state != SYSTEM_RUNNING)) {
        touch_softlockup_watchdog();
        return;
    }

    /* Wake up the high-prio watchdog task every second: */
    if (time_after(jiffies, touch_timestamp + HZ))
        wake_up_process(per_cpu(watchdog_task, this_cpu));

    /* Warn about unreasonable 10+ seconds delays: */
    if (time_after(jiffies, touch_timestamp + 10*HZ)) {
        per_cpu(print_timestamp, this_cpu) = touch_timestamp;

        spin_lock(&print_lock);
        printk(KERN_ERR "BUG: soft lockup detected on CPU#%d!\n",
            this_cpu);
        dump_stack();
        spin_unlock(&print_lock);
    }
}

/*
 * The watchdog thread - runs every second and touches the timestamp.
 */
static int watchdog(void * __bind_cpu)
{
    struct sched_param param = { .sched_priority = 99 };

    sched_setscheduler(current, SCHED_FIFO, ¶m);
    current->flags |= PF_NOFREEZE;

    /*
     * Run briefly once per second to reset the softlockup timestamp.
     * If this gets delayed for more than 10 seconds then the
     * debug-printout triggers in softlockup_tick().
     */
    while (!kthread_should_stop()) {
        set_current_state(TASK_INTERRUPTIBLE);
        touch_softlockup_watchdog();
        schedule();
    }

    return 0;
}

/*
 * Create/destroy watchdog threads as CPUs come and go:
 */
static int __cpuinit
cpu_callback(struct notifier_block *nfb, unsigned long action, void *hcpu)
{
    int hotcpu = (unsigned long)hcpu;
    struct task_struct *p;

    switch (action) {
    case CPU_UP_PREPARE:
        BUG_ON(per_cpu(watchdog_task, hotcpu));
        p = kthread_create(watchdog, hcpu, "watchdog/%d", hotcpu);
        if (IS_ERR(p)) {
            printk("watchdog for %i failed\n", hotcpu);
            return NOTIFY_BAD;
        }
          per_cpu(touch_timestamp, hotcpu) = jiffies;
          per_cpu(watchdog_task, hotcpu) = p;
        kthread_bind(p, hotcpu);
         break;
    case CPU_ONLINE:
        wake_up_process(per_cpu(watchdog_task, hotcpu));
        break;
#ifdef CONFIG_HOTPLUG_CPU
    case CPU_UP_CANCELED:
        if (!per_cpu(watchdog_task, hotcpu))
            break;
        /* Unbind so it can run.  Fall thru. */
        kthread_bind(per_cpu(watchdog_task, hotcpu),
                 any_online_cpu(cpu_online_map));
    case CPU_DEAD:
        p = per_cpu(watchdog_task, hotcpu);
        per_cpu(watchdog_task, hotcpu) = NULL;
        kthread_stop(p);
        break;
#endif /* CONFIG_HOTPLUG_CPU */
     }
    return NOTIFY_OK;
}

static struct notifier_block __cpuinitdata cpu_nfb = {
    .notifier_call = cpu_callback
};

__init void spawn_softlockup_task(void)
{
    void *cpu = (void *)(long)smp_processor_id();
    int err = cpu_callback(&cpu_nfb, CPU_UP_PREPARE, cpu);

    BUG_ON(err == NOTIFY_BAD);
    cpu_callback(&cpu_nfb, CPU_ONLINE, cpu);
    register_cpu_notifier(&cpu_nfb);

    atomic_notifier_chain_register(&panic_notifier_list, &panic_block);
}