Re: [PATCH v6 1/3] evtchn: slightly defer lock acquire where possible

[Jan Beulich <jbeulich@xxxxxxxx>]

On 28.05.2021 15:31, Roger Pau Monné wrote:
> I think I'm being extremely dull here, sorry. From your text I
> understand that the value returned by is_port_valid could be stale by
> the time the user reads it?
> 
> I think there's some condition that makes this value stale, and it's
> not the common case?

It's evtchn_destroy() running in parallel which can have this effect.

Jan