1.

I got a bug report 

BUG: spinlock recursion on CPU#0, hci0/1205 lock: dfae1008, .magic: dead4ead, .owner: hci0/1205, .owner_cpu: 0 [<c003ade0>] (unwind_backtrace+0x0/0xe0) from [<c0467c00>] (dump_stack+0x20/0x24) [<c0467c00>] (dump_stack+0x20/0x24) from [<c020d4b4>] (spin_bug+0x90/0xa4) [<c020d4b4>] (spin_bug+0x90/0xa4) from [<c020d624>] (_raw_spin_lock+0x54/0x168) [<c020d624>] (_raw_spin_lock+0x54/0x168) from [<c046aebc>] (_spin_lock+0x28/0x2c) [<c046aebc>] (_spin_lock+0x28/0x2c) from [<c03df364>] (hci_rx_task+0x268/0x498) [<c03df364>] (hci_rx_task+0x268/0x498) from [<c006e7b0>] (tasklet_action+0x78/0xcc) [<c006e7b0>] (tasklet_action+0x78/0xcc) from [<c006eeac>] (__do_softirq+0xfc/0x204) [<c006eeac>] (__do_softirq+0xfc/0x204) from [<c006f00c>] (irq_exit+0x58/0xac) [<c006f00c>] (irq_exit+0x58/0xac) from [<c0033080>] (asm_do_IRQ+0x80/0x98) [<c0033080>] (asm_do_IRQ+0x80/0x98) from [<c00 

2.

It is reported here in the linux kernel source code

lib/spinlock_debug.c

void _raw_spin_lock(spinlock_t *lock) { debug_spin_lock_before(lock); if (unlikely(!__raw_spin_trylock(&lock->raw_lock))) __spin_lock_debug(lock); debug_spin_lock_after(lock); } static inline void debug_spin_lock_before(spinlock_t *lock) { SPIN_BUG_ON(lock->magic != SPINLOCK_MAGIC, lock, "bad magic"); SPIN_BUG_ON(lock->owner == current, lock, "recursion"); SPIN_BUG_ON(lock->owner_cpu == raw_smp_processor_id(), lock, "cpu recursion"); } 

3. 

The reason is that, I did things like

spinlock(); printfk(...) spinunlock(); 

It is not allowed , because in printk() function, it will also call spinlock(), and that cause the recursion.