Hi, > > The scan_req struct contains a reference to which interface is scanning, > > so it should very well be possible to have > > > > phy0: > > wlan0: IFF_UP & scanning > > wlan1: IFF_UP & change MAC address all the time > > > > just like it's possible to change the MAC address when wlan1 *isn't* > > IFF_UP even if wlan0 is scanning, right? > > > > Indeed. But that is not what you were suggesting earlier with just > checking local->scanning. So if scan_req contains a wdev, then yes it > should be possible to compare the scan_req->wdev to the interface being > changed. Well, yes, but only because I was incrementally going from James's patch, which was checking that only. Similar with the other local-> things being checked here, btw, though in some cases it might be harder to actually determine which wdev is doing something and which isn't. > > No, this typically cannot be fixed, and it doesn't really make sense. > > The NIC cannot possibly do two scans at a time since it has only a > > single radio resource :-) > > So why is the scan request not per phy then? And should mac address > even affect the ongoing scan? Can we simply change it with the scan > ongoing? There are things that affect the scan from the interface, e.g. capability overrides, (extended) capabilities, the MAC address is used unless randomization is requested, etc. johannes
