Forwarding to Ethereal-dev
-----Original Message-----
From: Joe Marcus Clarke
On Fri, 2004-04-23 at 20:30, Joe Marcus Clarke wrote:
> I'm spawning tethereal-0.10.3 in Java on Windows 2000 SP4. I'm reading
> bytes from stdout, and writing them to a file. The resulting capture
> file is corrupt. It either claims to have been truncated in mid-packet,
> or it claims that one packet far exceeds the max capture size of 65535.
>
> I thought this must have something to do with my Java code (even though
> it works fine on Solaris and FreeBSD). So I took the same command line,
> and did a simple redirect to a file:
>
> tethereal -s 65535 -w - > outfile
>
> The resulting outfile has the same problem. Again, this command works
> fine on Solaris and FreeBSD (and I assume all flavors of UNIX). I've
> tried both WinPcap 2.3 and 3.0, and both exhibit the same behavior. If
> needed, I can produce one of the bad capture files. However, I think
> this is pretty reproduceable as it's now happened on two different
> Windows 2000 machines.
>
> I searched the archives, but didn't find anything relating to this.
> Ideally, what I'd like to be able to do is use -w <filename>, but
> Process.destroy() (in Java) calls TerminateProcess() on Windows, and
> this doesn't give tethereal a chance to flush its output buffer. If
> tethereal flushed after each packet that might do it, but it only seems
> to do that if the output file is stdout. Any advice would be greatly
> appreciated. Thanks.
I saw Richard Urwin's reply in the archives, but I was not copied. I
should have said I was not subscribed. In any event, it got me
thinking, and sure enough, the file descriptor is not opened in binary
mode if the output is stdout.
I believe the attached patch should do the trick, but I do not have
access to a C compiler on Windows to test for sure. I'd appreciate
comments. And please CC me as I am not subscribed to the list. Thanks.
Joe
>
> Joe
--
PGP Key : http://www.marcuscom.com/pgp.asc
Attachment:
file_access.c.diff
Description: Binary data