[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20090323085122.4c9d21f2@nehalam>
Date: Mon, 23 Mar 2009 08:51:22 -0700
From: Stephen Hemminger <shemminger@...ux-foundation.org>
To: Li Yang <leoli@...escale.com>
Cc: bridge@...ts.linux-foundation.org, netdev@...r.kernel.org,
Li Yang <leoli@...escale.com>
Subject: Re: [PATCH] net/bridge: use the maximum hard_header_len of ports
for bridging device
On Fri, 20 Mar 2009 17:04:29 +0800
Li Yang <leoli@...escale.com> wrote:
> The bridging device used a constant hard_header_len. This will cause
> headroom shortage for ports with additional hardware header. The patch
> makes bridging device to use the maximum value of all ports.
>
> Signed-off-by: Li Yang <leoli@...escale.com>
> ---
That ensures big enough header for locally generated packets, but
any drivers that need bigger headroom still must handle bridged packets
that come in with smaller space. When bridging packets, the skb comes
from the allocation by the receiving driver. Almost all drivers will
use dev_alloc_skb() which will allocate NET_SKB_PAD (16) bytes of
additional headroom. This is used to hold copy of ethernet header for
the bridge/netfilter code.
So your patch is fine as an optimization but a driver can not safely
depend on any additional headroom. The driver must check if there
is space, and if no space is available, reallocate and copy.
--
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