Re: [PATCH printk v4 03/39] printk: Prepare for SRCU console list protection

[Petr Mladek]

On Tue 2022-11-15 12:39:18, John Ogness wrote:
> On 2022-11-15, Petr Mladek <pmladek@xxxxxxxx> wrote:
> > On Mon 2022-11-14 17:34:56, John Ogness wrote:
> >> Provide an NMI-safe SRCU protected variant to walk the console list.
> >> 
> >> Note that all console fields are now set before adding the console
> >> to the list to avoid the console becoming visible by SCRU readers
> >> before being fully initialized.
> >> 
> >> This is a preparatory change for a new console infrastructure which
> >> operates independent of the console BKL.
> >> 
> >> --- a/kernel/printk/printk.c
> >> +++ b/kernel/printk/printk.c
> >> @@ -3254,7 +3303,7 @@ int unregister_console(struct console *console)
> >>  		return -ENODEV;
> >>  	}
> >>  
> >> -	hlist_del_init(&console->node);
> >> +	hlist_del_init_rcu(&console->node);
> >
> > This should be hlist_del_rcu() here so that a list walker that it just
> > processing this console could see the rest of the list.
> 
> hlist_del_init_rcu() does not prevent the rest of the list from being
> seen. It only unlinks @pprev for future iterators.

I see. I looked at hlist_del_init() instead of on the _rcu() variant.
hlist_del_init_rcu() is correct here.


> > It seems that hlist_unhashed() checks only node.pprev pointer
> > so that we even do not need to initialize it.
> >
> > We discussed this in v3 in the patch implementing
> > console_force_preferred_locked(), see
> > https://lore.kernel.org/r/Y20aBwNWT19YDeib@alley.
> > I forgot that unregister_console() might have the same problem.
> 
> For console_force_preferred_locked() it was about making sure that the
> console always appears registered, even though it is temporarily removed
> from the console list.
> 
> For unregister_console() we want it to be seen as removed from the
> console list.

Makes sense.

Best Regards,
Petr