[PATCH v5 02/17] mesh: factor out rsn initialization

Jouni Malinen j at w1.fi
Mon Jun 11 10:23:27 PDT 2018

On Thu, May 31, 2018 at 04:04:33PM -0700, Peter Oh wrote:
> On 05/31/2018 01:55 AM, Jouni Malinen wrote:
> >There is only a single call to wpas_mesh_init_rsn() after this patch
> >series. Could you please clarify why this factoring out is done?
> in the later patch, "use setup completion callback to complete mesh", this
> factored out function is moving to "wpas_mesh_complete_cb()"  which is not
> called before CAC is done, but only called after CAC is done. The original
> code which initializes RSN is a part of wpa_supplicant_mesh_init() that is
> called always regardless of CAC is done or not which leads auth failure when
> CAC is required.

That does not really answer my question. All that happens here is in
patch 3/15 where wpas_mesh_init_rsn() call is moved from
wpa_supplicant_mesh_init() into wpas_join_mesh(). No other users for
this functionality is added. In practice, patches 2 and 3 are just
moving this functionality into another function in the same file.

The commit message on this patch 2/17 seems to imply that there might be
multiple places where RSN initialization could be used ("RSN
initialization can be used in different phases if mesh initialization
and mesh join don't happen in sequence such as DFS CAC is done in
between, hence factor it out to help convering the case."). What is that
commit message talking about? Where are the other "different phases"?
That description does not seem to agree with what this patch series does
and there does not seem to be any real need for factoring this
functionality out into a separate function. It might be reasonable to do
so from other reasons to keep the functions shorter, but if that is the
case, the commit message better say so. Now it is just making one wonder
why this change is being done and what other "phases" there might be
missing from this series.

Jouni Malinen                                            PGP id EFC895FA

More information about the Hostap mailing list