netdev.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Pravin Shelar <pshelar@ovn.org>
To: Jiri Benc <jbenc@redhat.com>
Cc: Linux Kernel Network Developers <netdev@vger.kernel.org>,
	David Ahern <dsa@cumulusnetworks.com>
Subject: Re: [PATCH net-next] openvswitch: correctly fragment packet with mpls headers
Date: Mon, 3 Oct 2016 11:04:46 -0700	[thread overview]
Message-ID: <CAOrHB_Dh=OPUTbbN7QYo9Nz4cV99-k_fT4z7jSOEcnFEDKXQEg@mail.gmail.com> (raw)
In-Reply-To: <b0cb6efc9bfda8381af880f10fa6b3447f7d3d5d.1475512190.git.jbenc@redhat.com>

On Mon, Oct 3, 2016 at 9:33 AM, Jiri Benc <jbenc@redhat.com> wrote:
> If mpls headers were pushed to a defragmented packet, the refragmentation no
> longer works correctly after 48d2ab609b6b ("net: mpls: Fixups for GSO"). The
> network header has to be shifted after the mpls headers for the
> fragmentation and restored afterwards.
>
> Fixes: 48d2ab609b6b ("net: mpls: Fixups for GSO")
> Signed-off-by: Jiri Benc <jbenc@redhat.com>
> ---
>  net/openvswitch/actions.c | 26 +++++++++++++++++++++-----
>  1 file changed, 21 insertions(+), 5 deletions(-)
>
> diff --git a/net/openvswitch/actions.c b/net/openvswitch/actions.c
> index 4e03f64709bc..370b2ba3df4c 100644
> --- a/net/openvswitch/actions.c
> +++ b/net/openvswitch/actions.c
> @@ -62,7 +62,8 @@ struct ovs_frag_data {
>         struct vport *vport;
>         struct ovs_skb_cb cb;
>         __be16 inner_protocol;
> -       __u16 vlan_tci;
> +       u16 network_offset;     /* valid only if inner_protocol is set */
> +       u16 vlan_tci;
>         __be16 vlan_proto;
>         unsigned int l2_len;
>         u8 l2_data[MAX_L2_LEN];
> @@ -656,7 +657,6 @@ static int ovs_vport_output(struct net *net, struct sock *sk, struct sk_buff *sk
>
>         __skb_dst_copy(skb, data->dst);
>         *OVS_CB(skb) = data->cb;
> -       skb->inner_protocol = data->inner_protocol;
>         skb->vlan_tci = data->vlan_tci;
>         skb->vlan_proto = data->vlan_proto;
>
> @@ -666,6 +666,13 @@ static int ovs_vport_output(struct net *net, struct sock *sk, struct sk_buff *sk
>         skb_postpush_rcsum(skb, skb->data, data->l2_len);
>         skb_reset_mac_header(skb);
>
> +       if (data->inner_protocol) {
> +               skb->inner_protocol = data->inner_protocol;
> +               skb->inner_network_header = skb->network_header;
> +               skb_set_network_header(skb, data->network_offset);
> +       }
> +       skb_reset_mac_len(skb);
> +
>         ovs_vport_send(vport, skb);
>         return 0;
>  }
> @@ -684,7 +691,8 @@ static struct dst_ops ovs_dst_ops = {
>  /* prepare_frag() is called once per (larger-than-MTU) frame; its inverse is
>   * ovs_vport_output(), which is called once per fragmented packet.
>   */
> -static void prepare_frag(struct vport *vport, struct sk_buff *skb)
> +static void prepare_frag(struct vport *vport, struct sk_buff *skb,
> +                        u16 orig_network_offset)
>  {
>         unsigned int hlen = skb_network_offset(skb);
>         struct ovs_frag_data *data;
> @@ -694,6 +702,7 @@ static void prepare_frag(struct vport *vport, struct sk_buff *skb)
>         data->vport = vport;
>         data->cb = *OVS_CB(skb);
>         data->inner_protocol = skb->inner_protocol;
> +       data->network_offset = orig_network_offset;
>         data->vlan_tci = skb->vlan_tci;
>         data->vlan_proto = skb->vlan_proto;
>         data->l2_len = hlen;
> @@ -706,6 +715,13 @@ static void prepare_frag(struct vport *vport, struct sk_buff *skb)
>  static void ovs_fragment(struct net *net, struct vport *vport,
>                          struct sk_buff *skb, u16 mru, __be16 ethertype)
>  {
> +       u16 orig_network_offset = 0;
> +
> +       if (skb->inner_protocol) {
> +               orig_network_offset = skb_network_offset(skb);
> +               skb->network_header = skb->inner_network_header;
> +       }
> +
This is not correct way to detect MPLS packet. inner_protocol can be
set by any tunnel device for using tunnel offloads. So this would
break the fragmentation for encapsulated packets. How about using
eth_p_mpls() as done in do-output()?

  reply	other threads:[~2016-10-03 18:04 UTC|newest]

Thread overview: 8+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2016-10-03 16:33 [PATCH net-next] openvswitch: correctly fragment packet with mpls headers Jiri Benc
2016-10-03 18:04 ` Pravin Shelar [this message]
2016-10-04  8:24   ` Jiri Benc
2016-10-04  9:28     ` Jiri Benc
2016-10-04 16:53       ` Pravin Shelar
2016-10-04 16:59         ` Jiri Benc
2016-10-05  2:03           ` Pravin Shelar
2016-10-05 12:47             ` Jiri Benc

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to='CAOrHB_Dh=OPUTbbN7QYo9Nz4cV99-k_fT4z7jSOEcnFEDKXQEg@mail.gmail.com' \
    --to=pshelar@ovn.org \
    --cc=dsa@cumulusnetworks.com \
    --cc=jbenc@redhat.com \
    --cc=netdev@vger.kernel.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).