Comment # 3
on bug 8787
from Dominique Martinet
Actually one more question on what I changed:
the tree and everything inside are allocated with se_tree_create/se_alloc, but
the hashtable is created with gmalloc during the packet dissector init routine
and hashtable values/keys are also gmalloc'd
This means that if the capture is closed, the tree will be freed, then if it's
re-opened the hashtable will probably fetch the same pointers that no longer
point to valid data? (it will if the conversation index is going to be the
same)
Is there anything I could hook to that would let me properly clean the
hashtable at the same time se_alloc values are freed, at the end of the
capture?
(I'm assuming dissectors register_init_routine is only called once when
wireshark start, and not at every capture start. If there's a fresh init there
I guess I could reset the hashtables at this point.)
Anyway, thanks!
You are receiving this mail because:
- You are watching all bug changes.