On Fri, Jul 17, 2009 at 7:47 AM, Rémi Villé<[email protected]> wrote:
> 2009/7/17 Omprakash Gnawali <[email protected]>
>>
>> 2009/7/13 Mehmet Akif Antepli <[email protected]>:
>> > Hi Omprakash,
>> >
>> > About the inage field,
>> > In the code below, first there is the check for the neighbor table entry
>> > to
>> > update. Inage field is only decremented for the neighbor that sent
>> > beacon
>> > msg !
>> >
>> >   // update the inbound link quality by
>> >   // munging receive, fail count since last update
>> >   void updateNeighborTableEst(am_addr_t n) {
>> >     uint8_t i, totalPkt;
>> >     neighbor_table_entry_t *ne;
>> >     uint8_t newEst;
>> >     uint8_t minPkt;
>> >     minPkt = BLQ_PKT_WINDOW;
>> >     dbg("LI", "%s\n", __FUNCTION__);
>> >     for (i = 0; i < NEIGHBOR_TABLE_SIZE; i++) {
>> >       ne = &NeighborTable[i];
>> >       if (ne->ll_addr == n) {
>> >  if (ne->flags & VALID_ENTRY) {
>> >    if (ne->inage > 0)
>> >      ne->inage--;
>> >
>> > But i think inage field should be decremented when the following
>> > condition
>> > holds, isn't it?
>>
>> You caught a bug. This code became a bug when we changed from updating
>> all the entries in the link table to selectively updating the link
>> table. When all the entries were updated periodically, we would
>> decrement the age for all the entries and refresh the age only on the
>> entries that were heard from recently. We don't do such wholesome
>> periodic updates anymore. Thanks for catching this.
>>
>> - om_p
>
> What could be the bad effect of this bug? Some good neighbors could be
> evicted before a whole period?
> What is the correct code (maybe on cvs)?
>

inage did not have any impact on how the estimates were being
computed/updated so I removed the code. I just updated the CVS. You
can take a look and let me know if you still see some issues.

- om_p

_______________________________________________
Tinyos-help mailing list
[email protected]
https://www.millennium.berkeley.edu/cgi-bin/mailman/listinfo/tinyos-help

Reply via email to