Re: [PATCH v6 2/2] printk: Make printing of spin_dump() deferred to avoid a deadlock

[Byungchul Park]

On Mon, Mar 14, 2016 at 10:37:52AM +0900, Sergey Senozhatsky wrote:
> On (03/11/16 19:37), Byungchul Park wrote:
> [..]
> > +static void __spin_dump_deferred(raw_spinlock_t *lock, const char *msg)
> > +{
> > +	printk_func_t s;
> > +
> > +	s = this_cpu_read(printk_func);
> > +	this_cpu_write(printk_func, vprintk_deferred);
> > +
> > +	/*
> > +	 * To change printk_func, it must be in preempt disabled and irq
> > +	 * disabled. WARN_ON() should be called after the change because
> > +	 * the default printk_func which may be called from WARN_ON()
> > +	 * is prohibited in this context.
> > +	 */
> > +	WARN_ON(!preempt_count() || !irqs_disabled());
> > +	__spin_dump(lock, msg);
> > +
> > +	this_cpu_write(printk_func, s);
> > +
> > +	printk_pending_output();
> > +}
> >
> > +static void spin_dump(raw_spinlock_t *lock, const char *msg)
> > +{
> > +	if (unlikely(console_sem_spin_is_held()))
> > +		__spin_dump_deferred(lock, msg);
> > +	else
> > +		__spin_dump(lock, msg);
> > +}

Hello, Sergey

> 
> so can it be
> 
> vprintk_emit()
>  __spin_dump_deferred()
>   vprintk_deferred()
>    vprintk_emit()
>     __spin_dump_deferred()
      ^^^
      can be caused by raw_spin_lock(logbug_lock)

>      vprintk_deferred()

Yes, it can happen by raw_spin_lock(logbuf_lock) to print warning or error
message. Are you worrying about an infinite recursion?

1. In the case printing warning, eventually it can fill the buffer without
actual printing using console.

2. In the case printing error, the infinite recursion can be prevented by
debug_locks_off().

Therefore, no problem.

> 
> 
> or am I getting it wrong?

Please let me know if you have any opinion. :-)

Thanks,
Byungchul

> 
> 	-ss