[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20130120113825.759b4a58@nehalam.linuxnetplumber.net>
Date: Sun, 20 Jan 2013 11:38:25 -0800
From: Stephen Hemminger <stephen@...workplumber.org>
To: vyasevic@...hat.com
Cc: Michał Mirosław <mirqus@...il.com>,
netdev@...r.kernel.org, bridge@...ts.linux-foundation.org,
davem@...emloft.net, shemminger@...tta.com, mst@...hat.com,
shmulik.ladkani@...il.com
Subject: Re: [PATCH net-next V6 02/14] bridge: Add vlan filtering
infrastructure
On Sun, 20 Jan 2013 12:59:22 -0500
Vlad Yasevich <vyasevic@...hat.com> wrote:
> On 01/17/2013 08:57 PM, Michał Mirosław wrote:
> > 2013/1/16 Vlad Yasevich <vyasevic@...hat.com>:
> > [...]
> >> --- /dev/null
> >> +++ b/net/bridge/br_vlan.c
> > [...]
> >> +struct net_port_vlan *nbp_vlan_find(const struct net_port_vlans *v, u16 vid)
> >> +{
> >> + struct net_port_vlan *pve;
> >> +
> >> + /* Must be done either in rcu critical section or with RTNL held */
> >> + WARN_ON_ONCE(!rcu_read_lock_held() && !rtnl_is_locked());
> >> +
> >> + list_for_each_entry_rcu(pve, &v->vlan_list, list) {
> >> + if (pve->vid == vid)
> >> + return pve;
> >> + }
> >> +
> >> + return NULL;
> >> +}
> >
> > This looks expensive - it's O(n) with n = number of configured VLANs on a port.
> > And this is called for every packet. The bridge already has a hash of VLAN
> > structures found by br_vlan_find(). You could add a second bitmap there
> > (eg. ingres_ports[]) and check port's bit instead of walking the list.
> > You would use a bit more memory (64 bytes minus the removed list-head)
> > per configured VLAN but save some cycles in hot path.
> >
>
> Technically wouldn't even need another bitmap as an existing membership
> bitmap would cover this case. I did some profiling and the list is
> faster for 3 vlans per port. Hash is faster for more then 3 vlans.
>
> I can easily switch to hash if that is what others think.
>
> -vlad
Let's assume the people that really want this feature are using a lot
of vlan's. i.e n = 1000 or so. A bitmap is O(1). Any hash list would
incur a just a big memory penalty for the list head. In other words
a full bitmap is 4096 bits = 512 bytes. If you use hash list,
then the equivalent memory size would be only 64 list heads, therefore
a bitmap is a better choice than a hlist.
--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Powered by blists - more mailing lists