[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <CAM_iQpVY0Bs1nQDcZzzSKtGGFJ+qeZoXsjSHg4e8hg+MjWxtug@mail.gmail.com>
Date: Thu, 21 Jul 2016 09:50:00 -0700
From: Cong Wang <xiyou.wangcong@...il.com>
To: Jamal Hadi Salim <jhs@...atatu.com>
Cc: Linux Kernel Network Developers <netdev@...r.kernel.org>
Subject: Re: [Patch net-next 1/2] net_sched: move tc_action into tcf_common
On Thu, Jul 21, 2016 at 12:38 AM, Jamal Hadi Salim <jhs@...atatu.com> wrote:
> It is an involved change (and i was fond of the container_of()
> trickery;)).
>
> Looking at this patch I cant grok some parts - maybe i need to apply it
> when i am not in a hurry to see clearly.
> Mostly this around things like:
> ---
> const struct tc_action *a
> struct tcf_mirred *m = to_mirred(a)
> ---
>
> Essentially this reduces to
> struct tcf_mirred *m = (struct tcf_mirred *)a
>
> How would that work?
This works because we always put struct tc_action
as the first member of each specific action, e.g. tcf_mirred,
struct tcf_mirred {
struct tc_action common;
int tcfm_eaction;
int tcfm_ifindex;
int tcfm_ok_push;
struct net_device __rcu *tcfm_dev;
struct list_head tcfm_list;
};
We can put a container_of() if you want to relax this rule,
but we already use the same trick in many other places
in networking subsystem. ;)
> Do we still need the void *priv
No, I already removed it in this patch.
Thanks.
Powered by blists - more mailing lists