From mboxrd@z Thu Jan 1 00:00:00 1970 From: John Fastabend Subject: Re: [PATCH v2 04/14] vlan: Enable software emulation for vlan accleration. Date: Wed, 20 Oct 2010 20:32:12 -0700 Message-ID: <4CBFB43C.6000501@intel.com> References: <1287618974-4714-1-git-send-email-jesse@nicira.com> <1287618974-4714-5-git-send-email-jesse@nicira.com> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit Cc: David Miller , "netdev@vger.kernel.org" To: Jesse Gross Return-path: Received: from mga01.intel.com ([192.55.52.88]:10016 "EHLO mga01.intel.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1757083Ab0JUDcM (ORCPT ); Wed, 20 Oct 2010 23:32:12 -0400 In-Reply-To: <1287618974-4714-5-git-send-email-jesse@nicira.com> Sender: netdev-owner@vger.kernel.org List-ID: On 10/20/2010 4:56 PM, Jesse Gross wrote: > Currently users of hardware vlan accleration need to know whether > the device supports it before generating packets. However, vlan > acceleration will soon be available in a more flexible manner so > knowing ahead of time becomes much more difficult. This adds > a software fallback path for vlan packets on devices without the > necessary offloading support, similar to other types of hardware > accleration. > > Signed-off-by: Jesse Gross > --- > include/linux/netdevice.h | 14 +++++++++++--- > net/core/dev.c | 36 +++++++++++++++++++++++++++++++++--- > 2 files changed, 44 insertions(+), 6 deletions(-) > > diff --git a/include/linux/netdevice.h b/include/linux/netdevice.h > index 880d565..2861565 100644 > --- a/include/linux/netdevice.h > +++ b/include/linux/netdevice.h > @@ -2248,9 +2248,17 @@ static inline int skb_gso_ok(struct sk_buff *skb, int features) > > static inline int netif_needs_gso(struct net_device *dev, struct sk_buff *skb) > { > - return skb_is_gso(skb) && > - (!skb_gso_ok(skb, dev->features) || > - unlikely(skb->ip_summed != CHECKSUM_PARTIAL)); > + if (skb_is_gso(skb)) { > + int features = dev->features; > + > + if (skb->protocol == htons(ETH_P_8021Q) || skb->vlan_tci) > + features &= dev->vlan_features; > + > + return (!skb_gso_ok(skb, features) || > + unlikely(skb->ip_summed != CHECKSUM_PARTIAL)); > + } > + > + return 0; > } > > static inline void netif_set_gso_max_size(struct net_device *dev, > diff --git a/net/core/dev.c b/net/core/dev.c > index 4c3ac53..1bfd96b 100644 > --- a/net/core/dev.c > +++ b/net/core/dev.c > @@ -1694,7 +1694,12 @@ static bool can_checksum_protocol(unsigned long features, __be16 protocol) > > static bool dev_can_checksum(struct net_device *dev, struct sk_buff *skb) > { > - if (can_checksum_protocol(dev->features, skb->protocol)) > + int features = dev->features; > + > + if (vlan_tx_tag_present(skb)) > + features &= dev->vlan_features; > + > + if (can_checksum_protocol(features, skb->protocol)) > return true; > > if (skb->protocol == htons(ETH_P_8021Q)) { > @@ -1793,6 +1798,16 @@ struct sk_buff *skb_gso_segment(struct sk_buff *skb, int features) > __be16 type = skb->protocol; > int err; > > + if (type == htons(ETH_P_8021Q)) { > + struct vlan_ethhdr *veh; > + > + if (unlikely(!pskb_may_pull(skb, VLAN_ETH_HLEN))) > + return ERR_PTR(-EINVAL); > + > + veh = (struct vlan_ethhdr *)skb->data; > + type = veh->h_vlan_encapsulated_proto; > + } > + > skb_reset_mac_header(skb); > skb->mac_len = skb->network_header - skb->mac_header; > __skb_pull(skb, skb->mac_len); > @@ -1964,9 +1979,14 @@ static inline void skb_orphan_try(struct sk_buff *skb) > static inline int skb_needs_linearize(struct sk_buff *skb, > struct net_device *dev) > { > + int features = dev->features; > + > + if (skb->protocol == htons(ETH_P_8021Q) || vlan_tx_tag_present(skb)) > + features &= dev->vlan_features; > + > return skb_is_nonlinear(skb) && > - ((skb_has_frag_list(skb) && !(dev->features & NETIF_F_FRAGLIST)) || > - (skb_shinfo(skb)->nr_frags && (!(dev->features & NETIF_F_SG) || > + ((skb_has_frag_list(skb) && !(features & NETIF_F_FRAGLIST)) || > + (skb_shinfo(skb)->nr_frags && (!(features & NETIF_F_SG) || > illegal_highdma(dev, skb)))); > } > > @@ -1989,6 +2009,15 @@ int dev_hard_start_xmit(struct sk_buff *skb, struct net_device *dev, > > skb_orphan_try(skb); > > + if (vlan_tx_tag_present(skb) && > + !(dev->features & NETIF_F_HW_VLAN_TX)) { > + skb = __vlan_put_tag(skb, vlan_tx_tag_get(skb)); > + if (unlikely(!skb)) > + goto out; > + > + skb->vlan_tci = 0; > + } > + Nice set of patches! If we tag frames in dev_hard_start_xmit() can we consolidate the offload enabled and non-offloaded net_device_ops in 8021q. And then not tag in vlan_dev_hard_start_xmit? I'll post an example thinking out loud here. Thanks, John. > if (netif_needs_gso(dev, skb)) { > if (unlikely(dev_gso_segment(skb))) > goto out_kfree_skb; > @@ -2050,6 +2079,7 @@ out_kfree_gso_skb: > skb->destructor = DEV_GSO_CB(skb)->destructor; > out_kfree_skb: > kfree_skb(skb); > +out: > return rc; > } >