[PATCH v5 14/17] mesh: do not allow scan result to swap pri/sec
peter.oh at bowerswilkins.com
Tue Jun 12 11:07:42 PDT 2018
On 06/11/2018 09:58 AM, Jouni Malinen wrote:
> On Thu, May 31, 2018 at 05:54:33PM -0700, Peter Oh wrote:
> Could you please identify the exact limitations that "need this patch"?
> I'm trying to understand whether this is just trying to hide an issue
> that should really be fixed somewhere else.
The purpose of this patch is to avoid peer link failure due to swapping
primary and secondary channel.
But as I replied in "mesh: do not allow pri/sec channel switch" patch,
this exception may be not the right approach and could hide an existing
issue. But again this patch series is to make mesh point use DFS
channels, not to fix existing issue, so I'll remove this patch from the
series in next revision.
>> However this single patch can be dropped, since it's a kind of improving
>> reliability patch, not a functional patch. let me know if you want me to
>> drop it.
> Huh? Improving reliability? That does not match the previous claim that
> there are limitation that need this patch. So no, I'm certainly not
> taking this without a clear description of why it is needed.
This is my commit message of this patch. "Swapping between primary and
secondary channel will break mesh from joining, hence don't allow it."
swapping pri/sec channel does not alway happen which means mesh could be
still able to join each other without this patch if swapping pri/sec
doesn't happen after scan. But there are chances that swapping is
happening after scan due to neighbor environment. Improving reliability
here means improve peer link reliability by avoid such swap happening
which leads mesh peer link failure.
But again I understood your concern and will remove this patch from the
series and handle the swap pri/sec issue separately later.
More information about the Hostap