[olsr-dev] malloc thrashing / duplicate data structures ?

[Hannes Gredler]

hi,

i have done further code inspection on olsrd and wondered
if anybode shares my concern about malloc() thrashing and
duplicate tree insertions.

for example when doing the SPF computation the code

- repackages information that is already present from
  TC messages and generates a dedicated SPF tree.

apart from the insertion cost for *every* SPF run,
i am also a bit concerned about the level of
malloc() calls in the system.

IMO it would be desirable to have a malloc() call
only when a new neighbor comes up or you receive
an TC message for a new node and perform the
tree insertion only once using embedded (avl) tree
nodes.

or am i simply missing something ?

/hannes