UE_IP_ADDRESS without IE_UE_IP_ADDRESS_SD=1

Question

UE_IP_ADDRESS without IE_UE_IP_ADDRESS_SD=1

sothy opened this issue 4 years ago · comments

Hi,
My SMF sends PFCP session modification request with create_pdr.pdi.UEIP address.
This create_pdr has to detect the downlink flow based on UE ip address. In one setup, I didnt set UE as destination address. Anyhow, it was a problem of SMF. However, UPF sends a reply of PFCP session modification response with success. In this case, there is no entry in VRF 2 to process the downlink packet.
To be complete, UPF needs to send error message of "Semantically incorrect Information Element" in PFCP session modification response.

Am I right? It is not an urgent feature, but I prefer to keep an entry in the issue list.

yongshouwang · Answer 1 · Thu Mar 26 2020 10:26:13 GMT+0800 (China Standard Time)

Hello, I am a member of the oai project team and are working on your open source project featrure / 2001 / upf. During the research, I have the following questions:

The tests passed those PFCP protocols;
What effect did the test achieve? gtpu encap, decap, or forward traffic;
If the pfcp protocol is issued, user traffic can be normally forwarded, discarded, cached, etc. according to the PFCP protocol;
4.upf 2001 This project, how do you test, please give a test topology diagram;

thank you very much!

Andreas Schultz · Answer 2 · Thu Mar 26 2020 15:24:20 GMT+0800 (China Standard Time)

Hello, I am a member of the oai project team and are working on your open source project featrure / 2001 / upf. During the research, I have the following questions:

@yongshouwang stop spamming unrelated issues with your questions. Open a new issue (and only one) with you questions.

yongshouwang · Answer 3 · Fri Mar 27 2020 08:32:48 GMT+0800 (China Standard Time)

hello:
feature/2001/upf, how do you test, can far traffic, please give a test topology diagram;
thank you very much!

Andreas Schultz · Answer 4 · Mon Apr 20 2020 18:56:30 GMT+0800 (China Standard Time)

@sothy after considering this a bit more I can't see why this setting would be invalid.

The fact that this creates a combination of filters that does not permit traffic to pass is IMHO not relevant when checking the PFCP message. It is the responsibility for the PGW-C/SMF instance to make sure the message when considered as a whole is consistent.

I also don't see anything in specifications that would prohibit this combination of flags.
Especially TS 23.214, Table 5.2.2-1 does not limit the use of UE IP Address to any specific direction.
Within the PDI, the combination of Source Interface CORE and UE IP address set to DESTINATION seems to be valid. At the PDR or PDI level there is no notion of UL or DL, this is only assigned at the URR level when considering both PDR and FAR.

sothy · Answer 5 · Mon Apr 20 2020 19:49:55 GMT+0800 (China Standard Time)

@RoadRunnr yes you are right. It is responsibility of PGW-X/SMF.

Here UE IP address does not set DESTINATION based on your second paragraph . cant UPF reject or give error message?

Andreas Schultz · Answer 6 · Mon Apr 20 2020 20:31:31 GMT+0800 (China Standard Time)

Here UE IP address does not set DESTINATION based on your second paragraph . cant UPF reject or give error message?

On what basis? Source and Destination UE IPs are permitted in PDI. I can't see any information in the PDI that would make either one invalid.

sothy · Answer 7 · Mon Apr 20 2020 20:38:44 GMT+0800 (China Standard Time)

@RoadRunnr Make sure that we running two separate threads. For this one, I didnt post my show upf session message. When I do test again, I will post the message here later in the week.

In the other thread, I posted the correct upf session info. Since I remove the logic to use SDF, I'm forced to use pdr id into opaque field. so that I can make flow.