ANNOUNCEMENT: Live Wireshark University & Allegro Packets online APAC Wireshark Training Session
April 17th, 2024 | 14:30-16:00 SGT (UTC+8) | Online

Wireshark-users: Re: [Wireshark-users] [Wireshark-dev] radius dictionary -> segfault

From: Guy Harris <guy@xxxxxxxxxxxx>
Date: Thu, 10 Dec 2009 19:31:37 -0800
On Dec 9, 2009, at 6:29 PM, Mihaly Zachar wrote:

> I don't have it yet if you need I can do, but if you try it I'm almost
> sure that you can reproduce the problem, because we tried it on multiple
> PCs and on a Mac as well, with the same result.

I couldn't get it to crash - but Bill Meier found the problem; it was an uninitialized pointer structure member.  It might have been that, when I tried it, the chunk of data allocated for the data structure in question was freshly-allocated, and thus all zero, so that the pointer happened to be null.  He reproduced it on Windows, and the allocator might stuff bogus values into freshly-allocated memory, causing a crash.  It might also happen if enough memory gets allocated and freed, so that the chunk of data allocated for the data structure has junk in it, giving a bad pointer.

A fix has been checked in, initializing the pointer member in question to a null pointer:

	http://anonsvn.wireshark.org/viewvc/trunk/epan/dissectors/packet-radius.c?r1=31221&r2=31220&pathrev=31221

and is targeted for the 1.2.5 and 1.0.11 releases.