On Fri, 2017-06-30 at 11:48 +0200, Simon Wunderlich wrote:
> Our mac80211 patches have been upstreamed. However, the (experimental) 
> wpa_supplicant patches have not been adopted and I've reworked them a little 
> but they are not in a state to resubmit yet. So I don't expect the 
> wpa_supplicant portion to work yet, and you also may need to think of a good 
> way to have the mesh peers on a channel succession order.

For reference, an earlier version of the patchset was submitted as an
RFC in November with a bit of follow up discussion later:
 http://lists.infradead.org/pipermail/hostap/2016-November/036726.html
 http://lists.infradead.org/pipermail/hostap/2016-December/036741.html

There have been improvements since (e.g. solving the use after free
issue) that are not yet submitted. I cannot comment on the current
state of the patches though.

In general a major issue is making sure that all stations switch to the
same channel when a radar is detected. There is a simple mechanism in
the standard for this which to my knowledge is not implemented in
mac80211 at this point. But even with this in place you will likely
want a versatile mechanism in hostap for stations to switch to a common
channel should they end up on differing ones.

Regards,
Benjamin