[PATCH] ACS: always discard DFS channels when DFS isn't allowed
nico.escande at gmail.com
Sun May 8 03:42:44 PDT 2022
On Tue Apr 19, 2022 at 11:18 AM CEST, Nicolas Escande wrote:
> On Sun Apr 17, 2022 at 11:18 AM CEST, Jouni Malinen wrote:
> > Why is this removing the use of WPA_DRIVER_FLAGS_DFS_OFFLOAD completely
> > here? I'm not completely sure I understood what kind of a case could
> > have allowed ACS to select a DFS channel incorrectly. Maybe that is
> > something that should be addressed in the ACS implementation instead?
> > Please note that the DFS offload cases might use different style for
> > configuring the DFS operations, so it is not clear whether the
> > ieee80211h parameter is really applicable in all such cases and this
> > type of a change could result in breaking something that is already
> > deployed.
> So this code fetches the channels supported by the hardware and applies
> flags on it depending on whether or not we can use them and under what
> conditions. So for a regular driver (one that does not sets the
> WPA_DRIVER_FLAGS_DFS_OFFLOAD flag), when DFS is disabled by hostapd conf,
> any DFS channel will be flagged as disabled. This gets bypassed when the
> driver sets WPA_DRIVER_FLAGS_DFS_OFFLOAD.
> From what I understood, the DFS offload flag means that the driver
> itself should react to DFS events and handle things like CAC wait times.
> It shouldn't imply anything for the rest of the code, it shouldn't hide
> the fact that this channel shouldn't be used for example.
> This is problematic when doing ACS because we usually do not take into
> account any channel that is flagged as disabled. So this applies to DFS
> channels when the conf doesn't allow them.
> This lead me to cases where a conf that explicitely doesn't allow DFS
> channel (ieee80211h=0) to be used with ACS still enabled, still selected
> a DFS channel because the driver handles DFS itself with
> WPA_DRIVER_FLAGS_DFS_OFFLOAD flag, but doesn't bypass our ACS
> implementation with WPA_DRIVER_FLAGS_ACS_OFFLOAD.
> I understand that this could have implications but for me the "right"
> behaviour should be :
> - non usable channels should be flagged as such (so when conf doesn't
> allow them exclude them)
> - hostapd's ACS implem should select a "usable" channels from the list
> (unless ACS is offloaded with WPA_DRIVER_FLAGS_ACS_OFFLOAD in which
> case all bets are off)
> What do you think ?
> > --
> > Jouni Malinen PGP id EFC895FA
Any new input on this ?
More information about the Hostap