Re: [PATCH] lockdep: Speed up lockdep_unregister_key() with expedited RCU synchronization

[Breno Leitao]

On Fri, Mar 21, 2025 at 11:37:31AM +0100, Eric Dumazet wrote:
> On Fri, Mar 21, 2025 at 10:31 AM Breno Leitao <leitao@xxxxxxxxxx> wrote:
> >
> > lockdep_unregister_key() is called from critical code paths, including
> > sections where rtnl_lock() is held. For example, when replacing a qdisc
> > in a network device, network egress traffic is disabled while
> > __qdisc_destroy() is called for every network queue.
> >
> > If lockdep is enabled, __qdisc_destroy() calls lockdep_unregister_key(),
> > which gets blocked waiting for synchronize_rcu() to complete.
> >
> > For example, a simple tc command to replace a qdisc could take 13
> > seconds:
> >
> >   # time /usr/sbin/tc qdisc replace dev eth0 root handle 0x1: mq
> >     real    0m13.195s
> >     user    0m0.001s
> >     sys     0m2.746s
> >
> > During this time, network egress is completely frozen while waiting for
> > RCU synchronization.
> >
> > Use synchronize_rcu_expedited() instead to minimize the impact on
> > critical operations like network connectivity changes.
> 
> Running 'critical operations' with LOCKDEP enabled kernels seems a bit
> strange :)

Apologies, I meant to say that at Meta, we have certain service tiers
that can accommodate the performance trade-offs of running a "debug"
kernel. This kernel provides valuable feedback about its operations.

The aim is to identify any significant issues early on, rather than
having to debug "hard issues" (such as deadlocks, out-of-memory access,
etc) once the kernel is in production.

> Reviewed-by: Eric Dumazet <edumazet@xxxxxxxxxx>

Thanks for your support during this investigation,
--breno