Re: [RFC PATCH v2] Add rcu user eqs exception hooks for async page fault

[Frederic Weisbecker]

2012/11/28 Li Zhong <zhong@xxxxxxxxxxxxxxxxxx>:
> Thank you all for the review and education.
>
> Below are my current understandings and an update version. Would you
> please help to review it again and give your comments?
>
> Thanks, Zhong
>
> Now it seems to me that it is legal to call rcu_irq_exit/enter() without
> a matching rcu_irq_enter/exit() if the cpu is in non rcu idle state.
>
> As opposite, it is illegal to call rcu_irq_exit() without a matching
> rcu_irq_enter() if the cpu is in rcu idle state.
>
> But it seems legal to call rcu_irq_enter() without a matching
> rcu_irq_exit() if the cpu is in rcu idle state, regarding the
> dynticks_nesting value.   However, it seems not good to exit the rcu
> idle state, if we are actually entering into idle mode, so maybe it's
> better to call a matching rcu_irq_exit() before actually idle?
>
> As Frederic pointed out, we need a rcu_user_exit() to exit the user eqs
> (if we are in this state) in the beginning. But after some more
> thinking, I guess we might also need to call rcu_user_enter() after the
> waiting, if we get this page fault from user space. So maybe it's better
> to use rcu user eqs exception hooks here?

Makes sense.

>
> With rcu_user_exit() at the beginning, now rcu_irq_enter() only protects
> the cpu idle eqs, but it's not good to call rcu_irq_exit() after the cpu
> halt and the page ready.

Hmm, why is it not good?

>
> So I still want to remove it. And later if it shows that we really needs
> rcu somewhere in this code path, maybe we could use RCU_NONIDLE() to
> protect it. ( The suspicious RCU usage reported in commit
> c5e015d4949aa665 seems related to schedule(), which is not in the code
> path if we are in cpu idle eqs )

Yes but if rcu_irq_*() calls are fine to be called there, and I
believe they are because exception_enter() exits the user mode, we
should start to protect there right now instead of waiting for a
potential future warning of illegal RCU use.

>
> I think we still need Gleb's patch about the idle check in
> kvm_async_pf_task_wait(), and maybe another patch for the
> exit_idle()/enter_idle() issue.

Right.

Thanks.
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/