On 06/28/13 09:21, Evan Huus wrote:
The correct approach, I think, is to fix epan_cleanup() so that it
actually cleans up *everything* owned by epan, but then simply not
call it if we're about to exit. If there is some cleanup we do need to
do, then epan should provide an epan_cleanup_fast() or
epan_cleanup_minimal() or similar to call in those cases.
True, good points. Sounds like a good use case for another environment
variable (WIRESHARK_CLEAN_UP_ON_EXIT?) which we can set in our test
scripts (including the fuzz test script since that cleanup--freeing emem
currently--is necessary to detect canary corruption).