[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-Id: <20201211122612.869225-8-jonas@norrbonn.se>
Date: Fri, 11 Dec 2020 13:26:07 +0100
From: Jonas Bonn <jonas@...rbonn.se>
To: netdev@...r.kernel.org
Cc: pablo@...filter.org, laforge@...monks.org,
Jonas Bonn <jonas@...rbonn.se>
Subject: [PATCH net-next v2 07/12] gtp: use ephemeral source port
All GTP traffic is currently sent from the same source port. This makes
everything look like one big flow which is difficult to balance across
network resources.
>From 3GPP TS 29.281:
"...the UDP Source Port or the Flow Label field... should be set dynamically
by the sending GTP-U entity to help balancing the load in the transport
network."
Signed-off-by: Jonas Bonn <jonas@...rbonn.se>
---
drivers/net/gtp.c | 8 ++++++--
1 file changed, 6 insertions(+), 2 deletions(-)
diff --git a/drivers/net/gtp.c b/drivers/net/gtp.c
index 4a3a52970856..236ebbcb37bf 100644
--- a/drivers/net/gtp.c
+++ b/drivers/net/gtp.c
@@ -477,7 +477,7 @@ static int gtp_xmit_ip4(struct sk_buff *skb, struct net_device *dev)
__be32 saddr;
struct iphdr *iph;
int headroom;
- __be16 port;
+ __be16 sport, port;
int r;
/* Read the IP destination address and resolve the PDP context.
@@ -527,6 +527,10 @@ static int gtp_xmit_ip4(struct sk_buff *skb, struct net_device *dev)
return -EMSGSIZE;
}
+ sport = udp_flow_src_port(sock_net(pctx->sk), skb,
+ 0, USHRT_MAX,
+ true);
+
/* Ensure there is sufficient headroom. */
r = skb_cow_head(skb, headroom);
if (unlikely(r))
@@ -545,7 +549,7 @@ static int gtp_xmit_ip4(struct sk_buff *skb, struct net_device *dev)
iph->tos,
ip4_dst_hoplimit(&rt->dst),
0,
- port, port,
+ sport, port,
!net_eq(sock_net(pctx->sk),
dev_net(pctx->dev)),
false);
--
2.27.0
Powered by blists - more mailing lists