[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20150309092557.GB13498@vergenet.net>
Date: Mon, 9 Mar 2015 18:25:59 +0900
From: Simon Horman <simon.horman@...ronome.com>
To: Andy Zhou <azhou@...ira.com>
Cc: davem@...emloft.net, netdev@...r.kernel.org
Subject: Re: [patch net-next 3/4] net: refactor IPv4 and IPv6 fragmentation
APIs
On Mon, Mar 09, 2015 at 01:01:37AM -0700, Andy Zhou wrote:
> Both ip_fragment() and ip6_fragment() APIs assume skb has an
> attached netdev device, from which the MTU size can be derived.
> However, skbs incoming from OVS vports do not have an attached
> netdev device.
>
> This patch splits the original function into two parts: The core
> fragmentation logic is now provided by
> ip_fragment_mtu()/ip6_fragment_mut().
>
> The original APIs are kept as is. Their implementation now calls
> the new APIs. Any information derived from the attached netdev
> device is first derived in the original APIs and passed into the
> new APIs.
>
> In addition, The call back output function into the new APIs now
> accepts two arguments: a skb and an application specific pointer,
> which specifies additional information not directly associated
> with skb, such as OVS flow, to the output function.
I suggest handling that portion of the change separately.
It doesn't seem nearly as well defined as the rest of the change.
And cast below seems somewhat undesirable to me.
[snip]
> diff --git a/net/ipv4/ip_output.c b/net/ipv4/ip_output.c
> index a7aea20..b85fd34 100644
> --- a/net/ipv4/ip_output.c
> +++ b/net/ipv4/ip_output.c
> + return ip_fragment_mtu(skb, mtu, ll_rs, NULL, dev,
> + (int (*)(struct sk_buff *, void *output_arg))output);
> +}
> EXPORT_SYMBOL(ip_fragment);
--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Powered by blists - more mailing lists