releases/2.6.36.2/vlan-avoid-hwaccel-vlan-packets-when-vid-not-used.patch

   1 From jesse@nicira.com  Tue Dec  7 11:49:39 2010
   2 From: Jesse Gross <jesse@nicira.com>
   3 Date: Mon,  8 Nov 2010 13:23:01 -0800
   4 Subject: [stable] [PATCH 2.6.36 stable] vlan: Avoid hwaccel vlan packets when vid not used.
   5 To: stable@kernel.org
   6 Cc: netdev@vger.kernel.org, David Miller <davem@davemloft.net>
   7 Message-ID: <1289251381-6671-1-git-send-email-jesse@nicira.com>
   8
   9 From: Jesse Gross <jesse@nicira.com>
  10
  11 [This patch applies only to 2.6.36 stable.  The problem was introduced
  12 in that release and is already fixed by larger changes to the vlan
  13 code in 2.6.37.]
  14
  15 Normally hardware accelerated vlan packets are quickly dropped if
  16 there is no corresponding vlan device configured.  The one exception
  17 is promiscuous mode, where we allow all of these packets through so
  18 they can be picked up by tcpdump.  However, this behavior causes a
  19 crash if we actually try to receive these packets.  This fixes that
  20 crash by ignoring packets with vids not corresponding to a configured
  21 device in the vlan hwaccel routines and then dropping them before they
  22 get to consumers in the network stack.
  23
  24
  25 Reported-by: Ben Greear <greearb@candelatech.com>
  26 Tested-by: Nikola Ciprich <extmaillist@linuxbox.cz>
  27 Signed-off-by: Jesse Gross <jesse@nicira.com>
  28 Acked-by: David Miller <davem@davemloft.net>
  29 Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
  30
  31 ---
  32  net/8021q/vlan_core.c |    3 +++
  33  net/core/dev.c        |   10 ++++++++++
  34  2 files changed, 13 insertions(+)
  35
  36 --- a/net/8021q/vlan_core.c
  37 +++ b/net/8021q/vlan_core.c
  38 @@ -43,6 +43,9 @@ int vlan_hwaccel_do_receive(struct sk_bu
  39         struct net_device *dev = skb->dev;
  40         struct vlan_rx_stats     *rx_stats;
  41
  42 +       if (unlikely(!is_vlan_dev(dev)))
  43 +               return 0;
  44 +
  45         skb->dev = vlan_dev_info(dev)->real_dev;
  46         netif_nit_deliver(skb);
  47
  48 --- a/net/core/dev.c
  49 +++ b/net/core/dev.c
  50 @@ -2891,6 +2891,15 @@ static int __netif_receive_skb(struct sk
  51  ncls:
  52  #endif
  53
  54 +       /* If we got this far with a hardware accelerated VLAN tag, it means
  55 +        * that we were put in promiscuous mode but nobody is interested in
  56 +        * this vid. Drop the packet now to prevent it from getting propagated
  57 +        * to other parts of the stack that won't know how to deal with packets
  58 +        * tagged in this manner.
  59 +        */
  60 +       if (unlikely(vlan_tx_tag_present(skb)))
  61 +               goto bypass;
  62 +
  63         /* Handle special case of bridge or macvlan */
  64         rx_handler = rcu_dereference(skb->dev->rx_handler);
  65         if (rx_handler) {
  66 @@ -2927,6 +2936,7 @@ ncls:
  67                 }
  68         }
  69
  70 +bypass:
  71         if (pt_prev) {
  72                 ret = pt_prev->func(skb, skb->dev, pt_prev, orig_dev);
  73         } else {