Re: LOCKDEP: possible irq lock inversion dependency detected

[Thomas Gleixner <tglx@linutronix.de>]

On Tue, 2007-05-22 at 17:19 -0700, Sven-Thorsten Dietrich wrote:
> swapper/1 just changed the state of lock:                                       
>  (rtc_lock#2){-...}, at: [<ffffffff8085b185>] sbf_init+0x25/0xe0                
> but this lock was taken by another, hard-irq-safe lock in the past:             
>  (xtime_lock){+...}                                                             
>                                                                                 
> and interrupts could create inverse lock ordering between them.                 
>                                                                                 
>                                                                                 
> other info that might help us debug this:                                       
> no locks held by swapper/1.                                                     
>                                                                                 
> the first lock's dependencies:                                                  
> -> (rtc_lock#2){-...} ops: 2 {                                                  
>    initial-use  at:                                                             
>                         [<ffffffff8025d383>] mark_lock+0xf3/0x5b0               
>                         [<ffffffff8025e6f4>] __lock_acquire+0x664/0xf80         
>                         [<ffffffff8025f098>] lock_acquire+0x88/0xc0             
>                         [<ffffffff8047e2c5>] rt_spin_lock+0x35/0x40             
>                         [<ffffffff8020e162>] read_persistent_clock+0x22/0x1b0   
>                         [<ffffffff80867e86>] timekeeping_init+0x86/0x100        
>                         [<ffffffff808537af>] start_kernel+0x1bf/0x350           
>                         [<ffffffff80853179>] _sinittext+0x179/0x180             
>                         [<ffffffffffffffff>] 0xffffffffffffffff 

Hmm. That's the code in question:

void __init timekeeping_init(void)                                                                                                                    
{                                                                                                                                                     
        unsigned long flags;                                                                                                                          
        unsigned long sec = read_persistent_clock();                                                                                                  
                                                                                                                                                      
        write_seqlock_irqsave(&xtime_lock, flags);  

The rtc_lock is never taken inside the xtime_lock.

Looks like code reordering due to gcc extra magic. Which compiler ? 	

Thanks,

	tglx