Document microblaze ethernet behavior

[kiranshila]

If I'm not mistaken, when I have multiple 10 GbE devices in a model (SNAP), the MicroBlaze controller takes control of the first one alphabetically. This is both non-intuitive and undocumented, as far as I can tell.

I would be more than happy to write these docs if someone can point me in the direction of where they should go.

Additionally, I've been told that it's possible to "merge' the data streams of the MicroBlaze control line and the packetized data in the Simulink model. It would be nice if this were described / tutorialized as well.

[jack-h]

On Thu, 13 Oct 2022, 18:56 Kiran Shila, ***@***.***> wrote: If I'm not mistaken, when I have multiple 10 GbE devices in a model (SNAP), the MicroBlaze controller takes control of the first one alphabetically. This is both non-intuitive and undocumented, as far as I can tell.

Correct on all counts. I'm not sure where this is best documented, since there isn't really any documentation of any sort for the microblaze control beyond comments in header files in jasper_library/sw/jam. It seems like maybe there should be a whole section of the docs on control mechanisms, covering at least tcpborphserver and the microblaze tapcp server(?)

I would be more than happy to write these docs if someone can point me in the direction of where they should go. Additionally, I've been told that it's possible to "merge' the data streams of the MicroBlaze control line and the packetized data in the Simulink model. It would be nice if this were described / tutorialized as well.

I'm not sure exactly what this is asking. The simulink datastream and the ublaze control stream are always arbitrated internally to the ethernet core, and the microblaze interface is never exposed in simulink. Does that in any way provide useful information? Cheers Jack

…

— Reply to this email directly, view it on GitHub <#53>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAKWM7DKZIQJQMARGKLSE63WDBEMLANCNFSM6AAAAAAREQSFGY> . You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>

[kiranshila]

It seems like maybe there should be a whole section of the docs on control mechanisms, covering at least tcpborphserver and the microblaze tapcp server(?)

This makes a ton of sense. I'll poke around to see where this might go.

The simulink datastream and the ublaze control stream are always arbitrated internally to the ethernet core, and the microblaze interface is never exposed in simulink.

Ah ok. No problem then, I just thought I heard otherwise.

[jack-h]

On Thu, 13 Oct 2022, 19:35 Kiran Shila, ***@***.***> wrote: It seems like maybe there should be a whole section of the docs on control mechanisms, covering at least tcpborphserver and the microblaze tapcp server(?) This makes a ton of sense. I'll poke around to see where this might go. The simulink datastream and the ublaze control stream are always arbitrated internally to the ethernet core, and the microblaze interface is never exposed in simulink. Ah ok. No problem then, I just thought I heard otherwise.

I think the only subtlety here is that some of the ethernet cores have a simulink option to disable the cpu interface, which removes all the arbitration logic and thus saves a little resources. But, if you disable the cpu interface, the ublaze won't be able to use the port. (I don't know if it insists on trying to use the first alphabetical port even if this one has a disabled cpu interface but another port has an enabled interface)

…

— Reply to this email directly, view it on GitHub <#53 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAKWM7DTMUFXZCZZVNWSLGTWDBI6ZANCNFSM6AAAAAAREQSFGY> . You are receiving this because you commented.Message ID: ***@***.***>