[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20170203163003.00d794dc@cakuba.netronome.com>
Date: Fri, 3 Feb 2017 16:33:48 -0800
From: Jakub Kicinski <kubakici@...pl>
To: David Miller <davem@...emloft.net>
Cc: michael.chan@...adcom.com, netdev@...r.kernel.org,
Alexei Starovoitov <alexei.starovoitov@...il.com>
Subject: Re: [PATCH net-next v2 00/12] bnxt_en: Add XDP support.
On Fri, 03 Feb 2017 16:50:54 -0500 (EST), David Miller wrote:
> We don't set precedence by one driver saying "hey it's better to do
> things this way, forget what all the other drivers are doing." Rather
> we have a "discussion" about what the appropriate thing is to do and
> convert all the drivers only after a decision has been made.
Would making sure that if xdp_adjust_head() changes the starting offset
(and length) of frame and the program returns XDP_PASS - the stack will
see the changes made by xdp_adjust_head() fall under the same follow
the precedent rule? That is what Martin did for mlx4 and mlx5, and what
John did in virtio, but not what this patch set does (see my comment on
patch 11).
I should have double checked mlx4/mlx5 and made it clearer that there is
a precedent here in my review comment...
Powered by blists - more mailing lists