| Message ID | 1464043706-2843932-2-git-send-email-tom@herbertland.com |
|---|---|
| State | RFC, archived |
| Delegated to: | David Miller |
| Headers | show |
From: Tom Herbert <tom@herbertland.com> Date: Mon, 23 May 2016 15:48:20 -0700 > diff --git a/net/ipv4/fou.c b/net/ipv4/fou.c > index 5f9207c..96260c6 100644 > --- a/net/ipv4/fou.c > +++ b/net/ipv4/fou.c > @@ -807,13 +807,20 @@ int __fou_build_header(struct sk_buff *skb, struct ip_tunnel_encap *e, > u8 *protocol, __be16 *sport, int type) > { > int err; > + struct net *net; > Please order local variables from longest to shortest line. > err = iptunnel_handle_offloads(skb, type); > if (err) > return err; > > - *sport = e->sport ? : udp_flow_src_port(dev_net(skb->dev), > - skb, 0, 0, false); > + if (skb->dev) > + net = dev_net(skb->dev); > + else if (skb->sk) > + net = sock_net(skb->sk); This is getting rediculous. Why not just put the net namespace pointer into the tunnel encap object?
On Tue, May 24, 2016 at 3:01 PM, David Miller <davem@davemloft.net> wrote: > From: Tom Herbert <tom@herbertland.com> > Date: Mon, 23 May 2016 15:48:20 -0700 > >> diff --git a/net/ipv4/fou.c b/net/ipv4/fou.c >> index 5f9207c..96260c6 100644 >> --- a/net/ipv4/fou.c >> +++ b/net/ipv4/fou.c >> @@ -807,13 +807,20 @@ int __fou_build_header(struct sk_buff *skb, struct ip_tunnel_encap *e, >> u8 *protocol, __be16 *sport, int type) >> { >> int err; >> + struct net *net; >> > > Please order local variables from longest to shortest line. > >> err = iptunnel_handle_offloads(skb, type); >> if (err) >> return err; >> >> - *sport = e->sport ? : udp_flow_src_port(dev_net(skb->dev), >> - skb, 0, 0, false); >> + if (skb->dev) >> + net = dev_net(skb->dev); >> + else if (skb->sk) >> + net = sock_net(skb->sk); > > This is getting rediculous. Why not just put the net namespace pointer into > the tunnel encap object? Thanks, in this case it will be easier to just pass net in as an argument. Tom
diff --git a/net/ipv4/fou.c b/net/ipv4/fou.c index 5f9207c..96260c6 100644 --- a/net/ipv4/fou.c +++ b/net/ipv4/fou.c @@ -807,13 +807,20 @@ int __fou_build_header(struct sk_buff *skb, struct ip_tunnel_encap *e, u8 *protocol, __be16 *sport, int type) { int err; + struct net *net; err = iptunnel_handle_offloads(skb, type); if (err) return err; - *sport = e->sport ? : udp_flow_src_port(dev_net(skb->dev), - skb, 0, 0, false); + if (skb->dev) + net = dev_net(skb->dev); + else if (skb->sk) + net = sock_net(skb->sk); + else + return -EINVAL; + + *sport = e->sport ? : udp_flow_src_port(net, skb, 0, 0, false); return 0; } @@ -845,6 +852,14 @@ int __gue_build_header(struct sk_buff *skb, struct ip_tunnel_encap *e, void *data; bool need_priv = false; int err; + struct net *net; + + if (skb->dev) + net = dev_net(skb->dev); + else if (skb->sk) + net = sock_net(skb->sk); + else + return -EINVAL; if ((e->flags & TUNNEL_ENCAP_FLAG_REMCSUM) && skb->ip_summed == CHECKSUM_PARTIAL) { @@ -860,8 +875,7 @@ int __gue_build_header(struct sk_buff *skb, struct ip_tunnel_encap *e, return err; /* Get source port (based on flow hash) before skb_push */ - *sport = e->sport ? : udp_flow_src_port(dev_net(skb->dev), - skb, 0, 0, false); + *sport = e->sport ? : udp_flow_src_port(net, skb, 0, 0, false); hdrlen = sizeof(struct guehdr) + optlen;
With the implementation of Transports over UDP fou and gue build header may be called before skb->dev is set. This patch checks skb->dev and if it is not set then tries to get net from sock_net. Signed-off-by: Tom Herbert <tom@herbertland.com> --- net/ipv4/fou.c | 22 ++++++++++++++++++---- 1 file changed, 18 insertions(+), 4 deletions(-)