3 tincds, one down, pinging down machine from others loses connection

[ebina1]

Guus Sliepen wrote:

>
>It's hard for me to guess where the problem lies. Could you recompile
>tinc with CFLAGS = -g, run it with ulimit -c 10000 (for bash) or limit
>coredumpsize 10000 (for tcsh), kill -9 when it runs and then use gdb to
>get a backtrace? Something like this:
>
># ulimit -c 10000
># ./tincd -n <netname> -ddddd
>... wait until it hangs ....
># kill -9 `cat /var/run/tinc.netname.pid`
># gdb ./tincd core
>...
>(gdb) bt
>... backtrace follows here ...
>

I compiled with -g.  ran it under gdb with the -D option.
After it hung I sent it a kill -8.  The backtrace from gdb showed:

0x80530ca in lookup_subnet_ipv4 (address=0xbffff0a8) at subnet.c:285
285             if ((*address & p->net.ipv4.mask) == p->net.ipv4.address)
(gdb) bt
#0  0x80530ca in lookup_subnet_ipv4 (address=0xbffff0a8) at subnet.c:285
#1  0x80527c7 in route_ipv4 (packet=0xbffff0d4) at route.c:109
#2  0x80528d9 in route_outgoing (packet=0xbffff0d4) at route.c:151
#3  0x804f077 in handle_tap_input () at net.c:1323
#4  0x804f2af in main_loop () at net.c:1406
#5  0x8053ac0 in main (argc=5, argv=0xbffff874, envp=0xbffff88c) at 
tincd.c:372
(gdb)


Tinc:         Discussion list about the tinc VPN daemon
Archive:      http://mail.nl.linux.org/lists/
Tinc site:    http://tinc.nl.linux.org/