On Tue, Mar 26, 2002 at 04:36:37AM -0500, Adam wrote: > This or other way. How about the patch now?
Well, it has more global variables than I'd like; I'd prefer to have the state information be attached to the conversation (for example, adding an "eap_state_t" structure, which contains the conversation state, which would include both the EAP-TLS defragmentation state and the LEAP state, and keeping "leap_state_t" around for per-frame state, with two separate memory chunks for them). In addition, you might want to use the EAP ID variable to detect duplicate fragments and the like. This means that the "reassemble.h" code might not be what you'd use - but that was designed for IP-style reassembly, where you have a fragment ID field that's common to all fragments, and a byte offset, rather than an ID field that serves, in effect, as a fragmentation offset.
