I have 3.3.
Sometimes it just closes by itself.
Is there a log file?
I tried running it from a console but it detaches.

Install 3.3.1, whatever you are seeing, it is probably fixed in that release.

I can confirm the same issue with ktorrent-3.3.2 built from the ports on FreeBSD.
The last messages in ktorrent log file are always:

Code: Select all

ret != len
Authentication(S) to  : ok


Please note that the address is indeed empty in the last log line.
It appears that ktorrent just exists after that.

Perhaps related, perhaps not, I see that ktorrent regularly attempts to initiate connections to itself (my own IP address) and only "server-side" logic rejects those attempts.

Caught it under gdb:

Code: Select all

Program received signal SIGSEGV, Segmentation fault.
[Switching to Thread 80a9ef740 (LWP 100516)]
0x0000000803700108 in getnameinfo () from /lib/libc.so.7
(gdb) bt
#0  0x0000000803700108 in getnameinfo () from /lib/libc.so.7
#1  0x0000000800758736 in net::Socks::setSocksServerAddress () from /usr/local/kde4/lib/libbtcore.so.12
#2  0x0000000800758a30 in net::Socks::setSocksServerAddress () from /usr/local/kde4/lib/libbtcore.so.12
#3  0x0000000800758b0b in net::Socks::setSocksServerAddress () from /usr/local/kde4/lib/libbtcore.so.12
#4  0x00000008022d56e8 in QThreadPrivate::start () from /usr/local/lib/qt4/libQtCore.so.4


It looks like a peer disconnects before reverse-resolving kicks off and so NULL address gets passed somewhere unexpected.

Compile with debug information, so we can have a better backtrace.

I will do it later, if you still want it.
For now I've determined that the crash happens in libbtcore/net/reverseresolver.cpp in ReverseResolver::resolve method and it seems to be caused by NULL value returned from addr.address(). Adding a check for that condition seems to work around the issue.