[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <CAM_iQpU5MP546FjwdD_PtOUFiAKb_TiRBEY3b+Kt9mVcCY7MTQ@mail.gmail.com>
Date: Wed, 11 Oct 2017 09:34:51 -0700
From: Cong Wang <xiyou.wangcong@...il.com>
To: Jiri Pirko <jiri@...nulli.us>
Cc: Linux Kernel Network Developers <netdev@...r.kernel.org>,
David Miller <davem@...emloft.net>,
Jamal Hadi Salim <jhs@...atatu.com>,
Saeed Mahameed <saeedm@...lanox.com>, matanb@...lanox.com,
leonro@...lanox.com, mlxsw@...lanox.com
Subject: Re: [patch net-next 1/4] net: sched: make tc_action_ops->get_dev
return dev and avoid passing net
On Tue, Oct 10, 2017 at 2:19 PM, Jiri Pirko <jiri@...nulli.us> wrote:
> Tue, Oct 10, 2017 at 07:44:53PM CEST, xiyou.wangcong@...il.com wrote:
>>On Tue, Oct 10, 2017 at 12:30 AM, Jiri Pirko <jiri@...nulli.us> wrote:
>>> -static int tcf_mirred_device(const struct tc_action *a, struct net *net,
>>> - struct net_device **mirred_dev)
>>> +static struct net_device *tcf_mirred_get_dev(const struct tc_action *a)
>>> {
>>> - int ifindex = tcf_mirred_ifindex(a);
>>> + struct tcf_mirred *m = to_mirred(a);
>>>
>>> - *mirred_dev = __dev_get_by_index(net, ifindex);
>>> - if (!*mirred_dev)
>>> - return -EINVAL;
>>> - return 0;
>>> + return __dev_get_by_index(m->net, m->tcfm_ifindex);
>>
>>Hmm, why not just return m->tcfm_dev?
>
> I just follow the existing code. The change you suggest should be a
> separate follow-up patch.
Why?
Your goal is "make tc_action_ops->get_dev return dev and avoid passing net",
using m->tcfm_dev is simpler and could save you from adding a net pointer
to struct tcf_mirred too.
Powered by blists - more mailing lists