On Thu, Aug 08, 2019 at 01:13:46PM -0700, Reinette Chatre wrote:
> There is a locking order dependency between cpu_hotplug_lock and
> rdtgroup_mutex (cpu_hotplug_lock before rdtgroup_mutex) that has to be
> maintained. To do so in this flow you will find cpus_read_lock() in
> rdtgroup_schemata_write(), so quite a distance from where it is needed.
>
> Perhaps I should add a comment at the location where the lock is
> required to document where the lock is obtained?
Even better - you can add:
lockdep_assert_cpus_held();
above it which documents *and* checks too. :-)
--
Regards/Gruss,
Boris.
Good mailing practices for 400: avoid top-posting and trim the reply.
