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

Wireshark-dev: Re: [Wireshark-dev] File size limits on Linux and building for large file suppor

From: Ulf Lamping <ulf.lamping@xxxxxx>
Date: Wed, 27 Jun 2007 08:25:06 +0200
Stephen Fisher schrieb:
On Tue, Jun 26, 2007 at 08:27:03AM +0200, Ulf Lamping wrote:
You'll need to compile without libz, as libz is (optionally?) used to work with capture files and it will by default use 32 bit file offsets on machines which uses 32 bits for long values. So you can compile Wireshark/Tshark without libz (configure option?) or use a 64 bit machine that uses 64bits for "long" integers.
Would it be worth the effort to put some code in that would use the zlib reading and writing functions for uncompressed/compressed use < 2GB vs. normal reading and writing functions for non-compressed files of any size based on a user preference?
I don't know if that's possible, AFAIK the change for long file support is a compile time option, so you won't be able to use a user preference for it - unless you are working with some more magic, compiling two different modules and loading one of it based on that preference. However, that's a lot of work for a minor goal IMHO.

My preferred way would be to motivate the zlib people to support >2GB on "32bit long" machines "natively". However, I didn't spend any effort in that direction.

Regards, ULFL