From: "J.Smith"
Sent: Saturday, November 16, 2002 7:55 AM
Subject: Re: [Ethereal-dev] crash on reading tracefile ( NFS ? )
Thanks for the bugreport.
> >
> >When disabling rpc protocol I don't get any crash.
> >
>
> Same for me here. And more specifically, when disabling only the NFS
> protocol (RPC enabled again) I don't get a crash either. (At least on
> Ethereal 0.9.5, Win98. Haven't been able to check on my Ethereal 0.9.7
> installation yet.)
I tested your capture on Linux and Cygwin (on a w2k box) and both these
versions
can dissect the packets fine.
It seems the bug only triggers on windows boxes running the native version
of ethereal.
Can you try to run tethereal to dissect the packet (using the -V flag)
tethereal.exe -V -n -r crash.trace ?
It might tell us which packet that caused it do dump.
Maybe someone with a c compiler and debugger for windows can provide
something similar
to a GDB stacktrace. I cant read DrWatson logs.
>
> This seems to prove my initial suspicion that there is either a problem
with
> the NFS dissector of Ethereal, or a serious problem with my NFS
> servers/clients ;) Which is, by the way, the reason I started doing the
NFS
> traces in the first place, so I wouldn't be surprised if in the end it
> turned out that my NFS server/clients are sending 'illegal' or 'malformed'
> NFS requests/replies that cause Ethereal to choke here. (But that is just
a
> suspicion at this point).
No matter how illegal packets the boxes might send, it is always a bug if
ethereal crashes on them.