Re: [PATCH net-next 09/14] gtp: Allow configuring GTP interface as standalone
From: Tom Herbert <hidden>
Date: 2017-09-20 16:24:09
On Wed, Sep 20, 2017 at 9:07 AM, Andreas Schultz [off-list ref] wrote:
On 20/09/17 17:57, Tom Herbert wrote:quoted
On Wed, Sep 20, 2017 at 8:27 AM, Andreas Schultz [off-list ref] wrote:quoted
On 19/09/17 02:38, Tom Herbert wrote:quoted
Add new configuration of GTP interfaces that allow specifying a port to listen on (as opposed to having to get sockets from a userspace control plane). This allows GTP interfaces to be configured and the data path tested without requiring a GTP-C daemon.This would imply that you can have multiple independent GTP sockets on the same IP address.That is not permitted by the GTP specifications. 3GPP TS 29.281, section 4.3 states clearly that there is "only" one GTP entity per IP address.A PDP context is defined by the destination IP and the TEID. The destination port is not part of the identity of a PDP context.We are in no way trying change GTP, if someone runs this in a real GTP network then they need to abide by the specification. However, there is nothing inconsistent and it breaks nothing if someone wishes to use different port numbers in their own private network for testing or development purposes. Every other UDP application that has assigned port number allows configurable ports, I don't see that GTP is so special that it should be an exception.GTP isn't special, I just don't like to have testing only features in there when the same goal can be reached without having to add extra stuff. Adding code that is not going to be useful in real production setups (or in this case would even break production setups when enabled accidentally) makes the implementation more complex than it needs to be.
Well, you could make the same argument that allowing GTP to configured as standalone interface is a problem since GTP is only allowed to be with used with GTP-C. But, then we have something in the kernel that the community is expected to support, but requires jumping through a whole bunch of hoops just to run a simple netperf. The more that patches and features look like other things in the kernel that are already well established, the better the chances we can accept them and support them. It's probably a natural consequence of any large open source project, so sometimes it's worth the effort to add a few lines of complexity to get the benefits of community contribution and support. Tom