L2CAP: static size of BLE_L2CAP_COC_MTU inefficient?! about mynewt-nimble

[rymanluk]

@haukepetersen I guess you have issue when testing against 2 devices using Nimble, is this correct?

Anyway, I don't know why it is 100 but note that L2CAP can take os_mbuf not only from MSYS_1.
I would do it as MYNEWT_VAL and set it by default to MSYS_1_BLOCK_SIZE - headers as you suggested.

Could you do such a patch?

[haukepetersen]

I guess you have issue when testing against 2 devices using Nimble, is this correct?

yes, 5-10 devices running Nimble :-)

Anyway, I don't know why it is 100 but note that L2CAP can take os_mbuf not only from MSYS_1.

But did I understand correctly, that the MSYS buffers are ordered by their block size, with MSYS_1 having the smallest block size? So that would mean that setting it at least to MSYS_1_BLOCK_SIZE would indeed make sense.

Could you do such a patch?

I can certainly for the RIOT port, but I am not sure if I sufficient enough with mynewt to include such a fix into the rest of the code tree. But I guess I can at least give it a try :-)

[haukepetersen]

So here is my try: #315