[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <1471474686.29842.53.camel@edumazet-glaptop3.roam.corp.google.com>
Date: Wed, 17 Aug 2016 15:58:06 -0700
From: Eric Dumazet <eric.dumazet@...il.com>
To: John Fastabend <john.fastabend@...il.com>
Cc: xiyou.wangcong@...il.com, jhs@...atatu.com,
alexei.starovoitov@...il.com, brouer@...hat.com,
john.r.fastabend@...el.com, netdev@...r.kernel.org,
davem@...emloft.net
Subject: Re: [RFC PATCH 08/13] net: sched: support skb_bad_tx with lockless
qdisc
On Wed, 2016-08-17 at 12:36 -0700, John Fastabend wrote:
> Similar to how gso is handled skb_bad_tx needs to be per cpu to handle
> lockless qdisc with multiple writer/producers.
\
> @@ -1021,6 +1026,7 @@ err_out4:
> free_percpu(sch->cpu_bstats);
> free_percpu(sch->cpu_qstats);
> free_percpu(sch->gso_cpu_skb);
> + free_percpu(sch->skb_bad_txq_cpu);
This might be the time to group all these per cpu allocations to a
single one, to help data locality and decrease overhead of having XX
pointers.
Powered by blists - more mailing lists