two possible bugs

[Javier Cardona]

Marco,

On Thu, Feb 18, 2010 at 6:04 AM, Porsch, Marco <marco.porsch at siemens.com> wrote:
> Hi list,
>
> 1st) I think in hwmp_route_info_get a PREQ should only be processed, if it has the same SN an better metric (instead of better or equal).
>
> 2nd) In hwmp_prep_frame_process: u8 next_hop[ETH_ALEN] is written to, but never used. Should be used to buffer mpath->next_hop->sta.addr for use out of lock.

Both changes make sense to me.
Would you be willing to submit a patch to wireless-testing?
(http://linuxwireless.org/en/developers/Documentation/SubmittingPatches)

Thanks,

Javier

> Files taken from compat-wireless-2010-02-02.
>
> Regards
> Marco
>
>
> --- mesh_hwmp.c~        2010-02-18 14:44:47.000000000 +0100
> +++ mesh_hwmp.c 2010-02-18 14:45:32.000000000 +0100
> @@ -391,7 +391,7 @@ static u32 hwmp_route_info_get(struct ie
>                                if (SN_GT(mpath->sn, orig_sn) ||
>                                    (mpath->sn == orig_sn &&
>                                     action == MPATH_PREQ &&
> -                                    new_metric > mpath->metric)) {
> +                                    new_metric >= mpath->metric)) {
>                                        process = false;
>                                        fresh_info = false;
>                                }
> @@ -611,7 +611,7 @@ static void hwmp_prep_frame_process(stru
>
>        mesh_path_sel_frame_tx(MPATH_PREP, flags, orig_addr,
>                cpu_to_le32(orig_sn), 0, target_addr,
> -               cpu_to_le32(target_sn), mpath->next_hop->sta.addr, hopcount,
> +               cpu_to_le32(target_sn), next_hop, hopcount,
>                ttl, cpu_to_le32(lifetime), cpu_to_le32(metric),
>                0, sdata);
>        rcu_read_unlock();
> _______________________________________________
> Devel mailing list
> Devel at lists.open80211s.org
> http://open80211s.com/mailman/listinfo/devel
>



-- 
Javier Cardona
cozybit Inc.
http://www.cozybit.com