[olsr-dev] Changes and some bug fixes to olsrd

[Erik Tromp]

I assume that by "interference", you mean whether nodes running the new
(updated) OLSRD will still be backward compatible with nodes running the current
0.4.10 OLSRD.

We did a lot of testing on the bug fixes in our lab and only after successful
testing did I post the changes. However, we did not systematically test
interoperability of the fixes with 0.4.10 (although we did incidentally have a
mixed network "by accident").

Below is my opinion about interference to Freifunk by the changes I submitted,
point by point. Please note that this is just my initial estimation; to be sure
about things we'd have to do some testing.

* Added comment about 'Weight' and 'LinkQualityMult' parameters in
  sample .conf files.
--> No interference: no code, just comments

* Fix copying of old instead of new main IP address in chk_if_changed(...)
function
--> No interference. Local node issue. Also, I assume nodes in the Freifunk do
not change their IP address, so this should not be a problem in Freifunk.

* Fix link-set -> neighbor-set mismatch that occurs when a neighbor node
  changes its main IP address.
--> No interference. Local node issue. Also, I assume nodes in the Freifunk do
not change their IP address, so this should not be a problem in Freifunk.

* Fixed an inconsistency in the LQ MPR calculation, where a node would be
selected
  as MPR to a 2-hop neighbor even if the 2-hop neigbor could be reached in
  1 hop at a lower ETX.
--> Low risk of interference, assuming Freifunk uses a high number (7 ?) for
mpr_coverage .

* Fix a bug in link-quality handling of hello messages inside
  olsr_process_message_neighbors. This bug caused continuous flapping
  of MPR selection in certain situations.
--> Low risk of interference, assuming Freifunk uses a high number (7 ?) for
mpr_coverage .

* Fix duplicate entries of aliases in MID entries.
--> No interference. Local node issue. The 0.4.10 code does not seem to be
dependent on the presence of duplicates of entries of the same IP address in the
MID table.

* Remove no longer declared aliases from MID entries.
--> No interference, I assume nodes in the Freifunk do not change their IP
address (so no changes in the MID declarations). Also, I would think most nodes
in the Freifunk network have only one OLSR-interface (no MID entries).

* Fix: when link quality routing is enabled, HNA route calculation should be
based on
  link quality, not on hopcount.
--> No interference. Updates nodes will have better routes to HNA's than nodes
which are not updated. Both type of nodes will still be able to route to the
HNA, although the not-updated nodes may choose less optimal routes.

* Prevent two nodes that are advertising the same HNA from processing each
  other's HNA ending up in the two nodes routing the same (sub)net to each
  other in deadlock.
  To this end, added functions find_local_hna4_entry(...) and
  find_local_hna6_entry(...)
--> No interference. I assume that currently there are no two nodes in the
Freifunk advertising the same HNA.


----- Original Message ----- 
From: "Bernd Petrovitsch" <(spam-protected)>
To: "OLSR development" <(spam-protected)>
Sent: Monday, February 05, 2007 12:25 AM
Subject: Re: [olsr-dev] Changes and some bug fixes to olsrd


> On Mon, 2006-11-27 at 13:48 +0100, Erik Tromp wrote:
> > Dear Maintainers of the OLSRD code,
> >
> > I would like to propose a few fixes and enhancements to the current OLSR
code. The proposed changes are the diff file which can be
> > downloaded from http://home.tiscali.nl/levab001/olsrd-0.4.10.diff. The
diff's are with respect to the current version 0.4.10, as can
> > be downloaded at: http://www.olsr.org/releases/0.4/olsrd-0.4.10.tar.gz
> [...]
> > * httpinfo plugin:
> >   - make IP-addresses clickable
> >   - Show ETX value of a HNA route
> >   - Show number of MID aliases for each MID entry
> >   - Print LQ information only if LQ enabled
>
> Applied.
>
> > * Fix double definition MAX_IFS (16) and MAXIFS (8)
> >
> >
> > * Increased MAX_IFS to 32
>
> Will not apply.
> Reason: We already got rid of MAX_IFS (and I actually thought that I
> removed the last statically sized array of interfaces). So instead I
> will also remove that constant and move the nic_states variable into the
> struct interface (similar to the struct netbuf).
>
> > * Moved annoying printing of duplicate tables down to debug level 8.
>
> Applied.
>
> For the rest I'm personally not sure which hunks are related to which
> fix and which of the fixes must or should be included.
> >From the descriptions they all sound reasonable, however I'm not sure if
> and how they infere with the changes in FreiFunkFirmware.
> Opinions?
>
> Bernd
> -- 
> Firmix Software GmbH                   http://www.firmix.at/
> mobil: +43 664 4416156                 fax: +43 1 7890849-55
>           Embedded Linux Development and Services
>
>
> _______________________________________________
> olsr-dev mailing list
> (spam-protected)
> https://www.olsr.org/mailman/listinfo/olsr-dev
>