[PATCHv2 0/3] hostap: Add AP DFS support

Jouni Malinen j
Wed Aug 7 02:13:31 PDT 2013


On Mon, Aug 05, 2013 at 04:55:43PM +0200, Simon Wunderlich wrote:
> I'd like to bump after another month of silence - how can we proceed with this
> patchset?

Sorry about the delay. This has been one of the victims of travel,
vacation, and too good weather.. I think the only real blocker here is
the unconventional use of eloop. I was trying to come up with a proposal
on direction on how to make that more acceptable, but have not found
time to do that, so let me such provide some points about the issues for
now.

There needs to be possibility of running CAC while the main event loop
has already started and is running. There can be multiple interfaces
being controlled through the same eloop and one of the interfaces doing
long operations cannot block other interfaces from operating or even the
same interface from being able to reply to control interface commands.
As such, I would like to get rid of all the concurrent eloop instances;
there should be no eloop_run() or eloop_terminate() call added in the
patch.

The part that I did not find time for yet was to go through each case
here and propose a better approach. In general, I'd expect these
functions to register a timeout and return without initializing the
interface. Actual interface start would happen from a callback once the
DFS related operations have been completed.

I have the pending changes in a branch that is based on top of the
current master branch and the current snapshot is here:
http://w1.fi/p/dfs/. Like I mentioned previously, 2/2 is fine, so it is
just the eloop design in 1/2 that needs to be address for these to be
accepted.

-- 
Jouni Malinen                                            PGP id EFC895FA



More information about the Hostap mailing list