Re: PATCH? net/bridge/br_if.c: fix use after free in port_carrier_check()

[Oleg Nesterov <oleg@tv-sign.ru>]

On 02/20, David Howells wrote:
> 
> Oleg Nesterov <oleg@tv-sign.ru> wrote:
> 
> >  static void release_nbp(struct kobject *kobj)
> >  {
> >  	struct net_bridge_port *p
> >  		= container_of(kobj, struct net_bridge_port, kobj);
> > +
> > +	dev_put(p->dev);
> 
> Does this need to be done with the mutex held?

I think no. At least the current code does dev_put() without mutex held.

>                                                     And does anything actually pay
> attention to the refcount on dev?  I assume not...

I guess net/core/dev.c:netdev_wait_allrefs(), but not sure.

> Should you clear p->dev->br_port before calling dev_put()?

Looks like it is protected by RCU... Anyway the current code does the same.

> Looks reasonable.  I like it.
> 
> Acked-By: David Howells <dhowells@redhat.com>

Thanks! I'll re-send with a proper changelog later today.

Oleg.