Attention is currently required from: neels.
pespin has posted comments on this change. ( 
https://gerrit.osmocom.org/c/osmo-upf/+/28244 )

Change subject: add pfcp_endpoint
......................................................................


Patch Set 4:

(5 comments)

File include/osmocom/pfcp/pfcp_endpoint.h:

https://gerrit.osmocom.org/c/osmo-upf/+/28244/comment/ef9f36f5_883e1392
PS3, Line 51: struct osmo_pfcp_endpoint {
> you mean the name should change to "osmo_pfcp_endp"? […]
Given this probably ends up as a public API in a shared library I think this is 
precisely the time to pinpoint this kind of stuff.


https://gerrit.osmocom.org/c/osmo-upf/+/28244/comment/6274b4f3_4665da4a
PS3, Line 104:
> Do you mean add getters and setters? […]
You call it API bloat, I call it do not break ABI next time you add something 
new in eg. osmo_pfcp_endpoint.cfg

Those callbacks are only set once and then used internally, so adding a setter 
API to add those makes sense. This way you avoid ABI breakage.


File src/libosmo-pfcp/pfcp_endpoint.c:

https://gerrit.osmocom.org/c/osmo-upf/+/28244/comment/32653b2b_417a78dc
PS3, Line 122:  /* time() returns seconds since 1970 (UNIX epoch), but the 
recovery_time_stamp is coded in the NTP format, which is
> i'm pretty unsure about this timestamp coding, just know that wireshark ended 
> up showing the expecte […]
Grep for "ntp32" in there.


https://gerrit.osmocom.org/c/osmo-upf/+/28244/comment/0bdd32a2_08ff7d97
PS3, Line 182:  if (qe->m->is_response) {
> Let me rephrase your comment: […]
IMHO it also makes sense to have 2 different timer callbacks, one for requests 
and another for responses. You are unnecessarily still mixing stuff in the same 
code path here.


File src/libosmo-pfcp/pfcp_endpoint.c:

https://gerrit.osmocom.org/c/osmo-upf/+/28244/comment/4d7b0db4_2bed5e55
PS4, Line 268:  /* Slight optimization: Add sent requests to the start of the 
list: we will usually receive a response shortly
You say to the start of the queue, but you do add_tail in both.

Does PFCP actually retransmit responses actively? I don't think so?

Why do you have a common osmo_pfcp_endpoint_retrans_queue_add()? Again, it 
makes sense to have completely separate paths for responses and requests here, 
they are queued for totally different reasons.



--
To view, visit https://gerrit.osmocom.org/c/osmo-upf/+/28244
To unsubscribe, or for help writing mail filters, visit 
https://gerrit.osmocom.org/settings

Gerrit-Project: osmo-upf
Gerrit-Branch: master
Gerrit-Change-Id: Ic8d42e201b63064a71b40ca45a5a40e29941e8ac
Gerrit-Change-Number: 28244
Gerrit-PatchSet: 4
Gerrit-Owner: neels <[email protected]>
Gerrit-Reviewer: Jenkins Builder
Gerrit-CC: pespin <[email protected]>
Gerrit-Attention: neels <[email protected]>
Gerrit-Comment-Date: Fri, 10 Jun 2022 09:34:47 +0000
Gerrit-HasComments: Yes
Gerrit-Has-Labels: No
Comment-In-Reply-To: neels <[email protected]>
Comment-In-Reply-To: pespin <[email protected]>
Gerrit-MessageType: comment

Reply via email to