Web lists-archives.com

Re: [PATCH v6 1/2] pid: Replace pid bitmap implementation with IDR API




On Wed, 11 Oct 2017 20:54:32 -0400 Rik van Riel <riel@xxxxxxxxxxx> wrote:

> On Wed, 2017-10-11 at 15:37 -0700, Andrew Morton wrote:
> > On Wed, 11 Oct 2017 18:19:38 -0400 Gargi Sharma <gs051095@xxxxxxxxx>
> > wrote:
> > 
> > > This patch replaces the current bitmap implemetation for
> > > Process ID allocation. Functions that are no longer required,
> > > for example, free_pidmap(), alloc_pidmap(), etc. are removed.
> > > The rest of the functions are modified to use the IDR API.
> > > The change was made to make the PID allocation less complex by
> > > replacing custom code with calls to generic API.
> > 
> > I still don't understand the locking.  spin_lock_irq(&pidmap_lock) in
> > some places, rcu_read_lock() in others.
> > 
> > If the locking is indeed now correct, can we please get it fully
> > documented?  A comment at the pid_namespace.idr definition site would
> > suit.
> 
> Would you like me to send a follow-up patch to document the
> locking?

Sure.

> Documenting the locking on all the existing code, plus the
> new code, seems a little out of scope of an Outreachy
> internship...

I'm not referring to all the existing code!  Just this new
pid_namespace.idr's locking.  If it was protected by
spin_lock_irq(&pidmap_lock) everywhere then an explanation wouldn't be
needed.  But we have this oddball site where pidmap_lock isn't taken
but it uses rcu_read_lock() which is surprising to say the least. 
Readers could be forgiven for thinking that is a bug.