#6211 BLOC Update.: patch for support of per packet mesh ttl

[Zarro Boogs per Child]

#6211: patch for support of per packet mesh ttl
--------------------------+-------------------------------------------------
  Reporter:  mbletsas     |       Owner:  mbletsas                      
      Type:  enhancement  |      Status:  new                           
  Priority:  blocker      |   Milestone:  Update.2 (8.2.0)              
 Component:  wireless     |     Version:                                
Resolution:               |    Keywords:  libertas, ttl, mesh, Update.1?
  Verified:  0            |    Blocking:                                
 Blockedby:               |  
--------------------------+-------------------------------------------------

Comment(by dsaxena):

 Hi!

 Michail asked me to take a look at this issue and provide my input.
 I've read the patches, the thread archives, etc and here are my
 comments FWIW:

 I basically agree with the upstream folks on this one that providing
 a driver level sockopt is not the right way to go; however, given that
 we need to have something working for the August update I suggest
 a two prong approach:

    1. Go ahead and incoporate the existing patch into our kernel to allow
 applications/algorithms/etc to be developed as needed for the August
 update. (On a related note: it looks like olpc-2.6/stable has mesh_opts.c
 from Dec '07 in it but not the changes to tx.c and it is not being built).
 [[BR]]
    1. At the same time, we need to commit to implementing the more generic
 solution (SO_LL_TTL) with upstream and backport that into our kernel and
 applications when we feel confident with it. As has been pointed out,
 we're not going to get any reassurance before hand that SO_LL_TTL will be
 accepted and seeing as we're the only ones who want this ATM, we're going
 to have to write atleast some prelimenary proof-of-concept code and take
 the risk that it will need to go trough some revisions as upstream
 comments on it. Also, as Jim pointed out, we need to present the case as
 to why we need this in our application space. I don't see that in my
 perusal of the threads (though I may have missed something).  I don't know
 the network stack internals too well, but I can't imagine adding a new
 generic sockopt to be too complicated.

 This approach does require changing applications down the road, but it is
 just 1 call and from what I understand of the issue, there are not that
 many applications that actually will be using this. If there are and we're
 worried about 3-rd party applications/activities, we can maybe mitigate
 this further by providing a higher-level API.

-- 
Ticket URL: <http://dev.laptop.org/ticket/6211#comment:30>
One Laptop Per Child <http://laptop.org/>
OLPC bug tracking system