[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20140327081425.GA2479@macbook.localnet>
Date: Thu, 27 Mar 2014 08:14:26 +0000
From: Patrick McHardy <kaber@...sh.net>
To: Vlad Yasevich <vyasevic@...hat.com>
Cc: David Miller <davem@...emloft.net>, netdev@...r.kernel.org
Subject: Re: [PATCH v2 net] vlan: Mask off vlan acceleration features on vlan
device.
On Wed, Mar 26, 2014 at 08:53:12PM -0400, Vlad Yasevich wrote:
> On 03/26/2014 05:15 PM, Patrick McHardy wrote:
> > On Wed, Mar 26, 2014 at 05:10:38PM -0400, David Miller wrote:
> >> From: Vlad Yasevich <vyasevic@...hat.com>
> >> Date: Wed, 26 Mar 2014 16:28:31 -0400
> >>
> >>> Some drivers incorrectly assign vlan acceleration features to
> >>> vlan_features thus causing issues for Q-in-Q vlan configurations.
> >>> Prevent this once and for all by masking off acceleration features
> >>> for vlan devices until such time as we support stacked acceleration.
> >>>
> >>> Signed-off-by: Vlad Yasevich <vyasevic@...hat.com>
> >>
> >> So what if we do have chips that can offload Q-in-Q? Will we use
> >> a new flag?
> >
> > For 802.1ad that was the intention, using the NETIF_F_HW_VLAN_STAG_* flags.
> >
>
> But we can't really do that as we don't carry multiple vlan_tci in the
> skb. If the card advertises VLAN_STAG_*, then we'll offload that, but
> CTAG will have to be computed by software. More infrastructure is
> needed to support nested vlan offloads.
True, we'd need more room in the skb to store both tags for Q-in-Q.
--
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