Bill Pringlemeir wrote: > (gdb) bt > #0 0x0811bdf3 in pcache_outgoing_connection (n=0xd594444) at override.h:78 > #1 0x080fc20e in node_is_now_connected (n=0xd594444) at nodes.c:3213
It would be interesting to see the details of "n". So far I don't see how the assertion could trigger. > Any requests for stupid user tricks? Recompile with dmalloc? If anyone could check Gtk-Gnutella with valgrind again that might uncover some bugs. You can compile with -Doptimize='-DPROTECT_ATOMS' which can catch some bugs. There might be away to extend this so that at least some cases of access-after-free bugs can be caught as well. By the way, GCC 4.1 has finally offical support for the stack protector. Although I don't think any of the issues mentioned recently are stack-related, it can't hurt to take advantage of it. > I was going to do that, but we switch to SVN and it took a long time > for me to get SVN going with SSL. SVN snapshots are available now. These snapshots don't contain the .svn directories though because that makes a huge difference in size. -- Christian
pgptjuksVDaZg.pgp
Description: PGP signature
