Re: [RFC][PATCH v5 1/2] printk: Make printk() completely async

[Sergey Senozhatsky]

On (03/21/16 16:33), Jan Kara wrote:
[..]
> > > And by calling wake_up_process() under logbuf_lock, you actually introduce
> > > recursion issues for printk_deferred() messages which are supposed to be
> > > working from under rq->lock and similar. So I think you have to keep this
> > > section outside of logbuf_lock.
> > 
> > hm, in_sched (printk_deferred()) messages are printed by
> > irq work->wake_up_klogd_work_func(), not by wake_up_process()
> > from vprintk_emit(). or am I missing something?
> 
> Think of following:
> 
> some function
>   printk()
>     vprintk_emit()
>       spin_lock(&logbuf_lock);
>       ...
>       wake_up_process()
>         printk_deferred()
>           vprintk_emit() -> recursion on logbuf_lock

uh, indeed. I was more concerned about printk() calls that are
troublemakers and are already in wake_up_process() - spin_dump()s.
but yes, braking printk_deferred() in this case is a regression.
thanks for pointing that out. and also thanks to Byungchul.

	-ss

> Previously scheduler code was allowed to call printk_deferred() wherever it
> wanted...
> 
> So we are not supposed to call into the scheduler from under logbuf_lock...
> 
> 								Honza
> -- 
> Jan Kara <jack@xxxxxxxx>
> SUSE Labs, CR
>