[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <CACP96tTVoHKrVJD+nX9=PpV-tk1CQ31=9bxdFeGUvDaJUrmkWA@mail.gmail.com>
Date: Wed, 28 May 2014 13:18:55 -0400
From: sowmini varadhan <sowmini05@...il.com>
To: Eric Dumazet <eric.dumazet@...il.com>
Cc: Jamal Hadi Salim <jhs@...atatu.com>,
Niels Möller <nisse@...thpole.se>,
netdev <netdev@...r.kernel.org>, Jonas Bonn <jonas@...thpole.se>,
Jiri Pirko <jiri@...nulli.us>
Subject: Re: Scaling 'ip addr add' (was Re: What's the right way to use a
*large* number of source addresses?)
On Wed, May 28, 2014 at 10:48 AM, Eric Dumazet <eric.dumazet@...il.com> wrote:
> Following patch helps a lot
> (30 seconds to add 65536 addresses on lo interface instead of 52 seconds
> on my host)
I haven't had a chance to try out your patch yet (thanks for
the "dont do a del" caveat :-)) and I'm going to try it out now,
but reading your info makes me
wonder- in this case all the addresses that are being added
are static Ipv4 addrs, i.e., they are clearly IFA_F_PERMANENT
addresses, so check_lifetime() should not have too much work
to do (other than to examine the ifa_flags and continue). So
why is this making such a big difference?
Alternatively, afaict check_lifetime() is only interested in
an !IFA_F_PERMANENT. So maybe __inet_insert_ifa should
only do the {inet_hash_insert(..); /* reset check_lifetime_work */}
when it added a !PERMANENT address? Would this break
something else? (I realize it may need more changes elsewhere,
if the PERMANENT flag is modified on the fly, I've not checked
for those yet)
--Sowmini
--
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