[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-Id: <20150408.224404.1913719826015357860.davem@davemloft.net>
Date: Wed, 08 Apr 2015 22:44:04 -0400 (EDT)
From: David Miller <davem@...emloft.net>
To: ast@...mgrid.com
Cc: daniel@...earbox.net, tgraf@...g.ch, jiri@...nulli.us,
jhs@...atatu.com, netdev@...r.kernel.org
Subject: Re: [PATCH v3 net-next 2/2] tc: add 'needs_l2' flag to ingress
qdisc
From: Alexei Starovoitov <ast@...mgrid.com>
Date: Wed, 8 Apr 2015 16:26:15 -0700
> index 4cd5cf1aedf8..887033fbdfa6 100644
> --- a/net/sched/act_csum.c
> +++ b/net/sched/act_csum.c
> @@ -565,6 +565,8 @@ static struct tc_action_ops act_csum_ops = {
> .act = tcf_csum,
> .dump = tcf_csum_dump,
> .init = tcf_csum_init,
> + /* todo: fix pskb_may_pull in tcf_csum to not assume L2 */
> + .flags = TCA_NEEDS_L2,
> };
>
> MODULE_DESCRIPTION("Checksum updating actions");
All this module really cares about is where skb_network_header() is
for data accesses. And a simple offset would allow it to calculate
the pskb_may_pull() length argument correctly.
And it would therefore work without all of these expensive SKB share
check calls etc.
I really don't like your approach, and I'd rather you propagate the
offset into the BPF programs you generate. I bet you can do it in
an efficient way, and just haven't put in the effort to discover
how.
--
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