Wireshark-bugs: [Wireshark-bugs] [Bug 5284] new_packet_list: redissection + redraw crashes when

Date: Wed, 25 Apr 2012 14:57:03 -0700 (PDT)
https://bugs.wireshark.org/bugzilla/show_bug.cgi?id=5284

--- Comment #22 from Evan Huus <eapache@xxxxxxxxx> 2012-04-25 14:57:03 PDT ---
(In reply to comment #21)
> (In reply to comment #20)
> > The original option (per-edt memory pools) works correctly in all the cases
> > I've been able to come up with.
> 
> Will it really?
> > ep_dissect_new(edt1)
> > ...
> > ep_dissect_new(edt2)
> Stuff with edt1 allocating ep_mamory from pool2?
> > ...
> > ep_dissect_cleanup(edt2)
> pool 2 cleared
> > ...
> > ep_dissect_new(edt3)
> all ep allocs from pool3
> > ...
> > ep_dissect_cleanup(edt3)
> > ...
> > ep_dissect_cleanup(edt1)
> > 
> 
> Or am I missing something? the refcounting solution might bloat the memory
> usage somewhat but it seems safe(er). on the plus side is also that it shouldn't
> give any performance hit.

My understanding is that even if edt's get nested, the nesting happens outside
of epan_dissect_run() where the key is actually important. So something like:

01 ep_dissect_new(edt1)       key = NULL
02 ..ep_dissect_run(edt1)     key = edt1
03 ....ep_alloc()                 .
04 ..ep_dissect_new(edt2)         .
05 ....ep_dissect_run(edt2)   key = edt2
06 ......ep_alloc()               .
07 ..ep_dissect_cleanup(edt2)     .
08 ..ep_dissect_new(edt3)         .
09 ....ep_dissect_run(edt3)   key = edt3
10 ......ep_alloc()               .
11 ..ep_dissect_cleanup(edt3)     .
12 ep_dissect_cleanup(edt1)       .

So that even though we're sort of back in scope for edt1 at line 12 with
key=edt3, no more ep_allocs() happen because the ep_dissect_run() for that edt
(line 02) has already returned.

I could be wrong.

-- 
Configure bugmail: https://bugs.wireshark.org/bugzilla/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the assignee for the bug.
You are watching all bug changes.