Re: [PATCH] leds: trigger: use RCU to protect the led_cdevs list

[Johannes Berg]

Hi Pavel,

> > 
> > Cc: stable@xxxxxxxxxxxxxxx
> 
> I ... don't like idea of this going to stable.

OK, I guess that's fair. We've been running into the lockdep report for
a while - ever since we made the spinlock in iwlwifi no longer disable
IRQs all the time, which was meant to be a good thing ...

However, the scenario that could cause *real* deadlocks there is really
unlikely and requires four CPUs all doing the exactly right thing,
including a normally very rare *write lock* on the leddev_list_lock, so
I don't think in practice it'll be much of an issue.

johannes