Wireshark · Wireshark-dev: [Wireshark-dev] Problem with fuzz testing caused by exception with left an unclean tree

[Ulf Lamping <ulf.lamping@xxxxxx>]

Hi List!

While fuzz-testing, I "regularly" get an ASSERT in epan/proto.c line 2682:

       DISSECTOR_ASSERT(sibling->next == NULL);

When I remember correct, this is caused by a dissector catching an 
exception caused by "the fuzz" and continues to work with that packet. 
That left the tree unclean it seems.

I've commented this line out for some time now, without getting any 
other negative effect while fuzz-testing and elsewhere.

As I didn't digged deeper into that code: is it ok to simply remove this 
assert (with a comment, of cause) or will that have very ugly effects 
later? As this is a very rare case, it would be acceptable IMHO to have 
a minor memory leak or alike in that case.

Regards, ULFL