looking at these, i'm wondering whether the better move is to be doing the channel check inside the helper functions as well. i remember the node channel being .. weirdly sketchy thought. i'll go do some more deep diving on this before it lands.

In D47751#1088951, @adrian wrote:

looking at these, i'm wondering whether the better move is to be doing the channel check inside the helper functions as well. i remember the node channel being .. weirdly sketchy thought. i'll go do some more deep diving on this before it lands.

Let us know when you come to the final conclusion and we can do the review then; I'd rather do it once than thrice.

Yeah, it shouldn't take long.

ah I remember it now. ni->ni_chw has to do with the currently configured channel width, which can be changed via a HT_TXCHWIDTH action frame. The STA can request that the AP switch to 20MHz for transmitting to that station for now, even if the station is associated in 40MHz.

So when it's associated, ni_chw == the width of ni->ni_chan.

I think I'm going to add another helper function that returns true if the channel is 40MHz, the peer is 40MHz /and/ the currently selected width is 40MHz, so if someone requests a channel width change, it'll correctly start transmitting at 20MHz even though it's associated at 40MHz. (which is perfectly legal.)

Ugh now I remember this mess in more detail. :-) Stay tuned.

adrian added a child revision: D47754: amrr: remove duplicate logic, use ieee80211_ht_check_tx_ht().Tue, Nov 26, 1:53 AM

Staying tuned; please indicate when done.