[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <1460418539.6473.583.camel@edumazet-glaptop3.roam.corp.google.com>
Date: Mon, 11 Apr 2016 16:48:59 -0700
From: Eric Dumazet <eric.dumazet@...il.com>
To: Cong Wang <xiyou.wangcong@...il.com>
Cc: Lars Persson <lars.persson@...s.com>,
Lars Persson <larper@...s.com>,
Linux Kernel Network Developers <netdev@...r.kernel.org>,
Jamal Hadi Salim <jhs@...atatu.com>,
LKML <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH net v2] net: sched: do not requeue a NULL skb
On Mon, 2016-04-11 at 16:19 -0700, Cong Wang wrote:
> My point is, for example, in OOM case, we don't know processin
> more SKB would make it better or worse. Maybe we really need to
> check the error code to decide to continue to exit?
Really, given this bug has been there for a long time (v3.18 ???), I
doubt it matters.
Nothing can tell that following packets in the qdisc need any
transformation, and memory allocations.
So I would just fix the bug in the simplest way.
__qdisc_run() has all the checks needed to yield when needed
(if (quota <= 0 || need_resched())) , no need to add more complexity
there.
Powered by blists - more mailing lists