at86rf2xx: Basic mode and NETOPT_AUTOACK

[roberthartung]

1. Why is it not possible to choose the default RX_MODE (without AACK enabled)?

2. In at86rf2xx_netdev.c:_set(NETOPT_AUTOACK) the comment says /* don't set res to set netdev_ieee802154_t::flags */. However netdev_ieee802154_set never handles this cases which causes ifconfig <n> set -autoack to fail with an error message. What was the idea behind it?

[roberthartung]

@haukepetersen @miri64 You should be able to answer these questions ;)

[miri64]

@haukepetersen @miri64 You should be able to answer these questions ;)

I don't know much about this device driver, except how it interacts with the netdev API ;-), but I can try to answer it with my limited knowledge:

1. Why is it not possible to choose the default RX_MODE (without AACK enabled)?

I could name you some villages in Bohemia, if you like ;-).

2. In at86rf2xx_netdev.c:_set(NETOPT_AUTOACK) the comment says /* don't set res to set netdev_ieee802154_t::flags */. However netdev_ieee802154_set never handles this cases which causes ifconfig set -autoack to fail with an error message. What was the idea behind it?

Why would netdev_ieee802154_set handle that case? As far as I know auto-ack is pretty device specific, so this should be done in at86rf2xx_netdev.c:_set().

[roberthartung]

2. See drivers/at86rf2xx: Fix set/get of NETOPT_AUTOACK #8216

[PeterKietzmann]

@roberthartung #8213 was merged shortly after you opened this issue. Could you give a quick summary of the remaining problem in the current state? If I see it correctly, point 1. was not addressed but I don't quite understand the problem there.

[roberthartung]

@PeterKietzmann You mean #8216? ;-) 2. should be solved from my initial question, but 1. is still not possible, we are still always working in basic operation mode.

[PeterKietzmann]

we are still always working in basic operation mode.

This confuses me a bit. I was assuming we only provide the extended mode in RIOT?

[roberthartung]

You're absolutely right ... basic mode is not implemented. AACK is extended.

Btw. @PeterKietzmann do you mind if we implement the alternative frame buffer access mode? RIOT is currently writing the buffer first and then starting transmission which is slow. There is also a mode where you can write while preparing the send which is faster.

[PeterKietzmann]

Why would I mind :-)? It's just that I fear synchronization issues (e.g., sending 6LoWPAN fragments subsequently) and thus, I'd consider this a high impact feature which needs to be tested properly. Going through the release tests would be a good way to go. The cutoff date for high impact features is 26th March.

[PeterKietzmann]

This relates to ongoing work by @jia200x which is partly described in #11473 and #11483.

[PeterKietzmann]

@roberthartung #13798 got merged. an we close this issue?