[PATCH net-next] openvswitch: Do not set skb ignore_df

[PATCH net-next] openvswitch: Do not set skb ignore_df

[Pravin B Shelar]

Tunnel transmit code clear this bit, so setting ignore_df has
no effect.

Signed-off-by: Pravin B Shelar <pshelar@nicira.com>
---
 net/openvswitch/vport-geneve.c |    1 -
 net/openvswitch/vport-gre.c    |    2 --
 net/openvswitch/vport-vxlan.c  |    2 --
 3 files changed, 0 insertions(+), 5 deletions(-)

diff --git a/net/openvswitch/vport-geneve.c b/net/openvswitch/vport-geneve.c
index 484864d..8769be9 100644
--- a/net/openvswitch/vport-geneve.c
+++ b/net/openvswitch/vport-geneve.c
@@ -209,7 +209,6 @@ static int geneve_tnl_send(struct vport *vport, struct sk_buff *skb)
 	df = tun_key->tun_flags & TUNNEL_DONT_FRAGMENT ? htons(IP_DF) : 0;
 	sport = udp_flow_src_port(net, skb, 1, USHRT_MAX, true);
 	tunnel_id_to_vni(tun_key->tun_id, vni);
-	skb->ignore_df = 1;
 
 	err = geneve_xmit_skb(geneve_port->gs, rt, skb, fl.saddr,
 			      tun_key->ipv4_dst, tun_key->ipv4_tos,
diff --git a/net/openvswitch/vport-gre.c b/net/openvswitch/vport-gre.c
index d4168c4..ec57221 100644
--- a/net/openvswitch/vport-gre.c
+++ b/net/openvswitch/vport-gre.c
@@ -194,8 +194,6 @@ static int gre_tnl_send(struct vport *vport, struct sk_buff *skb)
 	df = tun_key->tun_flags & TUNNEL_DONT_FRAGMENT ?
 		htons(IP_DF) : 0;
 
-	skb->ignore_df = 1;
-
 	return iptunnel_xmit(skb->sk, rt, skb, fl.saddr,
 			     tun_key->ipv4_dst, IPPROTO_GRE,
 			     tun_key->ipv4_tos, tun_key->ipv4_ttl, df, false);
diff --git a/net/openvswitch/vport-vxlan.c b/net/openvswitch/vport-vxlan.c
index d7c46b3..dc6f625 100644
--- a/net/openvswitch/vport-vxlan.c
+++ b/net/openvswitch/vport-vxlan.c
@@ -175,8 +175,6 @@ static int vxlan_tnl_send(struct vport *vport, struct sk_buff *skb)
 	df = tun_key->tun_flags & TUNNEL_DONT_FRAGMENT ?
 		htons(IP_DF) : 0;
 
-	skb->ignore_df = 1;
-
 	src_port = udp_flow_src_port(net, skb, 0, 0, true);
 
 	err = vxlan_xmit_skb(vxlan_port->vs, rt, skb,
-- 
1.7.1

Re: [PATCH net-next] openvswitch: Do not set skb ignore_df

[Jesse Gross]

On Fri, Jan 2, 2015 at 1:27 PM, Pravin B Shelar <pshelar@nicira.com> wrote:
> Tunnel transmit code clear this bit, so setting ignore_df has
> no effect.
>
> Signed-off-by: Pravin B Shelar <pshelar@nicira.com>

Is it actually right for the bit to be cleared though? As discussed in
the previous thread on tunnel MTU handling, I think that fragmentation
should be done as a very last resort.

Re: [PATCH net-next] openvswitch: Do not set skb ignore_df

[Thomas Graf]

On 01/02/15 at 03:03pm, Jesse Gross wrote:
> On Fri, Jan 2, 2015 at 1:27 PM, Pravin B Shelar <pshelar@nicira.com> wrote:
> > Tunnel transmit code clear this bit, so setting ignore_df has
> > no effect.
> >
> > Signed-off-by: Pravin B Shelar <pshelar@nicira.com>
> 
> Is it actually right for the bit to be cleared though? As discussed in
> the previous thread on tunnel MTU handling, I think that fragmentation
> should be done as a very last resort.

Agreed. The source VXLAN VTEP should not fragment at all. I think we
need a separate skb_scrub_packet() for encaps at this point. Some of
the assumptions made for namespace handover do not apply.

Re: [PATCH net-next] openvswitch: Do not set skb ignore_df

[Pravin Shelar]

On Fri, Jan 2, 2015 at 12:03 PM, Jesse Gross <jesse@nicira.com> wrote:
> On Fri, Jan 2, 2015 at 1:27 PM, Pravin B Shelar <pshelar@nicira.com> wrote:
>> Tunnel transmit code clear this bit, so setting ignore_df has
>> no effect.
>>
>> Signed-off-by: Pravin B Shelar <pshelar@nicira.com>
>
> Is it actually right for the bit to be cleared though? As discussed in
> the previous thread on tunnel MTU handling, I think that fragmentation
> should be done as a very last resort.

ok, I missed the discussion, I will send another patch.