> Date: Fri, 27 Oct 2017 20:08:43 +0200 > From: Stefan Sperling <s...@stsp.name> > > On Fri, Oct 27, 2017 at 04:47:20PM +0200, Mark Kettenis wrote: > > I don't know, but with the ioctl removed, there is nothing that sets > > the IEEE80211_F_QOS flag anymore. That seems weird. > > This flag is just unused for now. > It is a left-over from unfinished 11n support code by damien. > But I expect that it will be useful. > > A-MPDU frames (sent with Tx aggregation) are QoS frames. > So we're not sending QoS frames yet, we are just receiving them. > The receiving code path can detect QoS by checking for QoS headers > in received frames, which works without a flag. > > But the flag is already wired up in some of the Tx logic, and I > expect the F_QOS flag will eventually be used as part of a Tx agg > implementation. It could be set by drivers in order to enable QoS > and thus advertise A-MPDU Tx support to net80211 (there is no good > reason for using QoS without 11n). The stack then knows that it > needs to supply QoS frame headers, and can provide a Tx subframe > re-ordering buffer for drivers for cases where firmware doesn't > provide one.
Thanks for the explanation!