Endpoint Registration Failure Due to OTA Cluster

[schrodingersket]

Commit 3f57b46931b42c06cec5a2a9c4c9d663103fd3bd broke endpoint registration for both the Wiser (0x0b) and ZLO (0x01) endpoints; I receive the following error for all commits after that on startup (which, notably, does not crash the application but does prevent reporting for clusters on endpoints which did not fully register):

eApp_ZLO_RegisterEndpoint 1

The error is printed from here: https://github.com/fairecasoimeme/ZiGate/blob/8d94e691cd1b7f653d3a3992f5a0806d442cde24/Module%20Radio/Firmware/src/ZiGate/Source/ZigbeeNodeControlBridge/app_zcl_event_handler.c#L183

After adding some log output, it looks like the first failure is when registering the OTA Server cluster in the Wiser endpoint.

[schrodingersket]

Just wanted to add some additional context to this. If I don't register all of the endpoints (Wiser, Livolo, Terncy, etc.), the application successfully starts, and the ZLO endpoint is successfully registered.

My initial thought is that this has to do with insufficient RAM when initializing all of the desired endpoints (particularly the OTA cluster); NXP recommends
sharing common cluster instances across endpoints in JN-UG-3115 to decrease RAM utilization. Are we doing this currently? If not, I think it would be a good idea to do so, since adding new endpoints for various manufacturers won't scale if we initialize a new endpoint for each one.

[badzz]

Are you sure the problem is new ?

[schrodingersket]

Yep. v3.1c did not have this issue; commit 3f57b46 fails on a clean build, but the one before (6000f73) works fine. I also tested ~10-12 commits in between 3f57b46 and v3.1d HEAD (8d94e69 at the time of this comment) with the same result.

[badzz]

But do you have a runtime.error or a compile error ?

[fairecasoimeme]

Hello, I already had this kind of issue. I updated the basic cluster file. Could you try to recompile and test ?

[schrodingersket]

I'm still seeing this as of 6ac632d. @badzz It's a runtime error.

Here's the UART output after a RESET:

u8NumberOfRegEndpoints : 0

 Status : 0

u8NumberOfRegEndpoints : 1

u8NumberOfRegEndpoints : 2

 Status : 0

u8NumberOfRegEndpoints : 3

 Status : 0

eApp_ZLO_RegisterEndpoint 1

tsCLD_Groups 4

tsCLD_GroupTableEntry 28Got bdb event 2

Got bdb event 1

Got stack event 2

[badzz]

ok issue is due to the number of timer available and needed . each endpoint with ota (5 currently) uses 1 timer. If I try to increase the number of timer by 1, it crashes on startup ..

[schrodingersket]

Is there a reason we couldn't share a single OTA cluster instance across all endpoints? If I understand correctly, there would then only be a single timer necessary, yeah?

[badzz]

yes , we are looking at this possibility indeed

On Wed, Jul 8, 2020 at 2:29 AM schrodingersket notifications@github.com wrote:

Is there a reason we couldn't share a single OTA cluster instance across all endpoints? If I understand correctly, there would then only be a single timer necessary, yeah?

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/fairecasoimeme/ZiGate/issues/310#issuecomment-655209575, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABR5UYMKBRFOKKOKGTVFVBLR2O4X3ANCNFSM4OODV4UQ .

[schrodingersket]

Excellent. Happy to help with that if you like. I'm looking at potentially adding a couple of endpoints for some manufacturer-specific clusters (air quality data from Climax and Develco, in particular), so I'll be in and around those modules anyway.

[badzz]

https://github.com/fairecasoimeme/ZiGate/pull/315

[schrodingersket]

Closing in favor of #315