[PATCH 2/3] P2P: always re-select oper channel if not hard coded

[Arik Nemtsov]

On Tue, Sep 4, 2012 at 8:52 PM, Arik Nemtsov <arik at wizery.com> wrote:
> Since the operating channel is randomly set to 1/6/11 on init, which is
> always in the channel intersection, we were effectively ignoring the set
> of p2p preferred channels.
> Fix this by trying to get the best channel we can, unless the user hard
> coded the operating channel in the configuration file. Fall back to the
> initial randomly selected channel if a better one cannot be chosen.
>
> Signed-hostap: Arik Nemtsov <arik at wizery.com>
> ---
>  src/p2p/p2p_go_neg.c |   17 ++++++++++++++++-
>  1 file changed, 16 insertions(+), 1 deletion(-)
>
> diff --git a/src/p2p/p2p_go_neg.c b/src/p2p/p2p_go_neg.c
> index 031b3a1..e1ba466 100644
> --- a/src/p2p/p2p_go_neg.c
> +++ b/src/p2p/p2p_go_neg.c
> @@ -391,6 +391,20 @@ static void p2p_reselect_channel(struct p2p_data *p2p,
>         }
>
>         /*
> +        * Try to see if the original channel is in the intersection. If
> +        * so, no need to change anything, as it already contains some
> +        * randomness.
> +        */
> +       if (p2p_channels_includes(intersection, p2p->op_reg_class,
> +                                 p2p->op_channel)) {
> +               wpa_msg(p2p->cfg->msg_ctx, MSG_DEBUG,
> +                       "P2P: Using original operating class and channel "
> +                       "(reg_class %u channel %u) from intersection",
> +                       p2p->op_reg_class, p2p->op_channel);
> +               return;
> +       }
> +
> +       /*
>          * Fall back to whatever is included in the channel intersection since
>          * no better options seems to be available.
>          */
> @@ -639,7 +653,8 @@ void p2p_process_go_neg_req(struct p2p_data *p2p, const u8 *sa,
>                                 wpa_hexdump(MSG_DEBUG, "P2P: channels",
>                                             c->channel, c->channels);
>                         }
> -                       if (!p2p_channels_includes(&intersection,
> +                       if (!p2p->cfg->cfg_op_channel ||
> +                           !p2p_channels_includes(&intersection,

There's a similar fix for p2p_process_go_neg_resp() which I forgot
here. But let's see if Jouni is ok with this approach before sending a
v2.

Arik