[Devel] Re: [RFC][PATCH 06/14] Populate pid_nrs list with entry for init-pid-ns

[Herbert Poetzl]

On Wed, Mar 21, 2007 at 03:23:28AM -0600, Eric W. Biederman wrote:
> sukadev at us.ibm.com writes:
> 
> > From: Cedric Le Goater <clg at fr.ibm.com>
> > Subject: [RFC][PATCH 06/14] Populate pid_nrs list with entry for init-pid-ns
> >
> > Create/destroy the pid->pid_nrs list - when allocating/freeing  a struct pid.
> > The pid_nrs list contains just a single struct pid_nr for now, (corresponding
> > to init-pid-ns). 
> >
> > To enable finding a process based on any of its pid_t values, replace the
> > use struct pid_nr, rather than struct pid, in the pid_hash table. 
> >
> > Finally, reimplement find_pid() and pid_nr() based on the pid_nrs list.
> >
> > Changelog:
> > 	- [Serge Hallyn's comment]: Add comments on what pid->lock protects
> > 	  and that pid->nr will eventually go away.
> > 	- [Eric Biederman and containers list comments]: Reworked patches
> > 	  to drop support for unsharing pid namespace and to replace
> > 	  struct pid with struct pid_nr in the pid_hash table.
> >
> > Signed-off-by: Cedric Le Goater <clg at fr.ibm.com>
> > Signed-off-by: Sukadev Bhattiprolu <sukadev at us.ibm.com>
> > ---
> >  include/linux/pid.h |    9 ------
> >  kernel/pid.c | 73 ++++++++++++++++++++++++++++++++++------------------
> >  2 files changed, 50 insertions(+), 32 deletions(-)
> >
> 
> > @@ -242,50 +247,70 @@ struct pid_nr *alloc_pid_nr(struct pid_n
> >  	 */
> >  	pid_nr->pid = pid;
> >
> > +	spin_lock_irq(&pidmap_lock);
> > +	hlist_add_head_rcu(&pid_nr->pid_chain, &pid_hash[pid_hashfn(nr)]);
> > +	spin_unlock_irq(&pidmap_lock);
> > +
> 
> Hmm.  I think we really need to hash in some of the bits of the
> pid_namespace so processes with identical pids don't get put on
> the same hash chain.  It isn't critical but it is a good idea.

yes, can't hurt to distribute that a little, was
already thinking about something like this:

#define pid_hashfn(nr) hash_long((unsigned long)(nr ^ space), pidhash_shift)

as the hash_long should distribute any changes
quite evenly ...

best,
Herbert

> Eric
> _______________________________________________
> Containers mailing list
> Containers at lists.linux-foundation.org
> https://lists.linux-foundation.org/mailman/listinfo/containers
_______________________________________________
Containers mailing list
Containers at lists.linux-foundation.org
https://lists.linux-foundation.org/mailman/listinfo/containers