[support] MR Flag Prob

[Ben McCarthy]

Ok thanks, I thought it was unlikely that this hadn't been experienced by
anyone before. I actually also thought I was using 0.4 before I checked so
I'll probably have to get round to upgrading again at some point :(

Cheers,

Ben

> -----Original Message-----
> From: Romain KUNTZ [mailto:kuntz at lsiit.u-strasbg.fr]
> Sent: 03 June 2009 10:07
> To: Ben McCarthy
> Cc: 'Support ML'
> Subject: Re: MR Flag Prob
> 
> Hi Ben,
> 
> That's a nice catch... but a bit too late ;-) The latest nepl patch
> for UMIP 0.4 already corrects this, and uses an uint16_t to store the
> mob_rtr variable. So you did it the right way.
> 
> Cheers,
> romain
> 
> On 2009/06/02, at 15:54, Ben McCarthy wrote:
> 
> > Hi Romain,
> >
> > I have actually just found the problem. The device I am using is big
> > endian
> > and therefore the value of IP6_MH_BU_MR gets set to 0x0400. However,
> > mob_rtr
> > in the home_addr_info struct is only set to be a uint8_t so 400 is
> > too big
> > for it. Setting it to a uint16_t solves the problem.
> >
> > I am only using umip 0.3 so this may have been solved in 0.4, if not
> > what do
> > you think is the best permanent fix? Change the mob_rtr variable or
> > change
> > the values of IP6_MH_BU_ETC to all fit in uint8_t?
> >
> > Cheers,
> >
> > Ben
> >
> >> -----Original Message-----
> >> From: Romain KUNTZ [mailto:kuntz at lsiit.u-strasbg.fr]
> >> Sent: 02 June 2009 14:05
> >> To: Ben McCarthy
> >> Cc: Support ML
> >> Subject: Re: MR Flag Prob
> >>
> >> Hi Ben,
> >>
> >> I've added the support ML as architecture issues have been discussed
> >> recently there.
> >>
> >> On 2009/06/01, at 19:38, Ben McCarthy wrote:
> >>> I just stuck a check in to process_first_home_bu after the line:
> >>>
> >>> bule->flags = (IP6_MH_BU_HOME | IP6_MH_BU_ACK |
> >>>                       hai->lladdr_comp | hai->mob_rtr);
> >>>
> >>> I put in:
> >>>
> >>> if(bule->flags & IP6_MH_BU_MR)
> >>>                        Print..
> >>>              Else
> >>>                        ..
> >>>
> >>> And it appears that IP6_MH_BU_MR isn't set at this stage. Could this
> >>> be anything to do with the difference in endianess between x86 and
> >>> mips?
> >>>
> >>> Are there any other checks you can recommend I perform?
> >>
> >> Could you check the value of "hai->mob_rtr" in the above code? It is
> >> supposed to be set to IP6_MH_BU_MR when the "IsMobRtr" flag is
> >> enabled
> >> in the config file of the MR (it is set in src/gram.y).
> >>
> >> Cheers,
> >> romain
> >
> >
> >