[Intel-wired-lan] [next, S23, 06/13] i40e/i40evf: tunnels can be generic
Jeff Kirsher
jeffrey.t.kirsher at intel.com
Wed Jan 20 22:13:35 UTC 2016
On Tue, 2016-01-19 at 19:54 -0800, Alexander Duyck wrote:
> On 12/09/2015 03:50 PM, Hay, Joshua A wrote:
> > From: Jesse Brandeburg <jesse.brandeburg at intel.com>
> >
> > Since the i40e driver now supports VxLAN, GRE, and soon
> > to be Geneve tunnels, the driver can just note that an skb
> > contains a tunneled packet generically, instead of calling
> > out VXLAN specifically. The tunnel set up we do for hardware
> > is almost always the same for all tunnel types.
> >
> > This specifically enables ATR/Flow Director on GRE packets
> > which increases performance/scaling. Without this patch
> > GRE RX packets end up on a random RSS assigned queue.
> >
> > Signed-off-by: Jesse Brandeburg <jesse.brandeburg at intel.com>
> > Change-ID: Ie8e205603654910d2f718592cbef8132ae0719e4
> > Tested-by: Andrew Bowers <andrewx.bowers at intel.com>
> > ---
> > drivers/net/ethernet/intel/i40e/i40e_txrx.c | 7 ++++---
> > drivers/net/ethernet/intel/i40e/i40e_txrx.h | 2 +-
> > drivers/net/ethernet/intel/i40evf/i40e_txrx.c | 2 +-
> > drivers/net/ethernet/intel/i40evf/i40e_txrx.h | 2 +-
> > 4 files changed, 7 insertions(+), 6 deletions(-)
> >
> > diff --git a/drivers/net/ethernet/intel/i40e/i40e_txrx.c
> b/drivers/net/ethernet/intel/i40e/i40e_txrx.c
> > index 28d2943..ba79f67 100644
> > --- a/drivers/net/ethernet/intel/i40e/i40e_txrx.c
> > +++ b/drivers/net/ethernet/intel/i40e/i40e_txrx.c
> > @@ -2005,7 +2005,7 @@ static void i40e_atr(struct i40e_ring
> *tx_ring, struct sk_buff *skb,
> > if (!(tx_flags & (I40E_TX_FLAGS_IPV4 | I40E_TX_FLAGS_IPV6)))
> > return;
> >
> > - if (!(tx_flags & I40E_TX_FLAGS_VXLAN_TUNNEL)) {
> > + if ((tx_flags & I40E_TX_FLAGS_TUNNEL)) {
> > /* snag network header to get L4 type and address */
> > hdr.network = skb_network_header(skb);
> >
>
> So this patch has a major bug right here. Specifically it is
> disabling
> ATR for everything since the logic got flipped an you are checking
> the
> outer headers for tunneled frames, and the inner headers for
> non-tunneled frames.
>
> There is another bug a bit further down from here where the variable
> protocol is used which means it is outer header only and as such
> breaks
> any cases where you have v4 over v6 or v6 over v4.
Thanks Alex for finding this, Jesse is working on a revised patch right
now.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 819 bytes
Desc: This is a digitally signed message part
URL: <http://lists.osuosl.org/pipermail/intel-wired-lan/attachments/20160120/f6e716a6/attachment.asc>
More information about the Intel-wired-lan
mailing list