Re: [PATCH v2 rcu 0/8] NMI-safe SRCU reader API

From: Paul E. McKenney
Date: Tue Oct 18 2022 - 11:24:26 EST


On Tue, Oct 18, 2022 at 12:39:40PM +0206, John Ogness wrote:
> Hi Paul,
>
> On 2022-09-29, "Paul E. McKenney" <paulmck@xxxxxxxxxx> wrote:
> > This RFC series provides the second version of an NMI-safe SRCU reader
> > API in the guise of srcu_read_lock_nmisafe() and
> > srcu_read_unlock_nmisafe().
>
> I would like to post a new series for printk that will rely on the
> NMI-safe reader API. From the feedback of this series it appears that
> the RFC v2 is an acceptable starting point. How should we proceed?

Currently, I have this series on -rcu branch srcunmisafe.2022.10.03a,
with Frederic's patches on branch "dev". I will be producing a shiny
new branch with your fix and Frederic's debug later today, Pacific Time.
With luck, based on v6.1-rc1.

> Will Frederic be sending a patch for the extra safety checks using
> srcu_nmi_flags? Are you planning on posting a new version? Should I
> include this or another version within my upcoming series?

I will be incorporating these commits from Frederic:

6558b914fc4e ("srcu: Warn when NMI-unsafe API is used in NMI")
5dc788627109 ("srcu: Explain the reason behind the read side critical section on GP start")
54a118fce487 ("srcu: Debug NMI safety even on archs that don't require it")

Are there other patches I should be applying?

> Thanks for all your help on this!

And looking forward to the new improved printk()!

Thanx, Paul