[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <CAM_iQpXvynRVFW3e=sojVp1njBhq4GyTY1zgdUMoD2KqicBvqw@mail.gmail.com>
Date: Tue, 7 Jun 2016 16:39:20 -0700
From: Cong Wang <xiyou.wangcong@...il.com>
To: Eric Dumazet <eric.dumazet@...il.com>
Cc: Eric Dumazet <edumazet@...gle.com>,
"David S . Miller" <davem@...emloft.net>,
netdev <netdev@...r.kernel.org>,
Jamal Hadi Salim <jhs@...atatu.com>,
John Fastabend <john.fastabend@...il.com>,
Kevin Athey <kda@...gle.com>,
Xiaotian Pei <xiaotian@...gle.com>
Subject: Re: [PATCH net-next 2/2] net: sched: do not acquire qdisc spinlock in
qdisc/class stats dump
On Mon, Jun 6, 2016 at 4:53 PM, Eric Dumazet <eric.dumazet@...il.com> wrote:
> Really sounds overkill to care about these, as probably no one needs to
> get a 'consistent view of all these counters in a snapshot'.
>
> Even as of today, the qlen/backlog pair is wrong. No one ever used these
> values in an SNMP agent.
>
> Note that qlen/backlog is changed both by enqueue/dequeue, so the
> seqcount protection would not work.
I see.
>
> With the percpu stats thing, stats can not be fetched in a 'consistent'
> way.
Fair enough.
Thanks!
Powered by blists - more mailing lists