[Opendnssec-user] ods-signerd 1.4.10 crash

[Berry A.W. van Halderen]

On 09/15/2016 12:04 AM, Havard Eidnes wrote:
>> stack trace:
>>
>> Core was generated by `ods-signerd'.
>> Program terminated with signal 11, Segmentation fault.
>> #0  0x000000000042a45a in netio_dispatch ()
>> (gdb) where
>> #0  0x000000000042a45a in netio_dispatch ()
>> #1  0x000000000040df3e in xfrhandler_start ()
>> #2  0x000000000040e26e in xfrhandler_thread_start ()
>> #3  0x00007f7ff560b3ae in ?? () from /usr/lib/libpthread.so.1
>> #4  0x00007f7ff6075e90 in ___lwp_park50 () from /usr/lib/libc.so.12
>> #5  0x00007f7ff4400000 in ?? ()
>> #6  0x00007f7ff7ff14c0 in ?? ()
>> #7  0x0000000111110001 in ?? ()
>> #8  0x0000000033330003 in ?? ()
>> #9  0x0000000000000000 in ?? ()
>> (gdb) 
>> (gdb) x/i netio_dispatch
>>    0x42a383 <netio_dispatch>:   push   %r15
>> (gdb)
> 
> With debug symbols:
> 
> Program terminated with signal 11, Segmentation fault.
> #0  netio_dispatch (netio=0x7f7ff7b2a0c0, timeout=<optimized out>, sigmask=0x0)
>     at wire/netio.c:250
> 250                 relative.tv_sec = handler->timeout->tv_sec;
> (gdb) p handler
> $1 = (netio_handler_type *) 0x7f7fe300b2b8
> (gdb) p $->timeout
> $2 = (struct timespec *) 0x7522203031203031
> (gdb) p handler->timeout->tv_sec
> Cannot access memory at address 0x7522203031203031
> (gdb) where
> #0  netio_dispatch (netio=0x7f7ff7b2a0c0, timeout=<optimized out>, sigmask=0x0)
>     at wire/netio.c:250
> #1  0x000000000040df3e in xfrhandler_start (xfrhandler=0x7f7ff7b76090)
>     at daemon/xfrhandler.c:133
> #2  0x000000000040e26e in xfrhandler_thread_start (arg=<optimized out>)
>     at daemon/engine.c:255
> #3  0x00007f7ff560b3ae in ?? () from /usr/lib/libpthread.so.1
> #4  0x00007f7ff6075e90 in ___lwp_park50 () from /usr/lib/libc.so.12
> #5  0x00007f7ff4400000 in ?? ()
> #6  0x00007f7ff7ff14c0 in ?? ()
> #7  0x0000000111110001 in ?? ()
> #8  0x0000000033330003 in ?? ()
> #9  0x0000000000000000 in ?? ()
> (gdb) 
> 
> Looks like either "use after free" or "someone scribbled
> somewhere they should not".

Thank you for reporting it.  I suspect this might be a problem that the
signer still is busy with a transfer, or wants to do a transfer for a
zone that is no longer there.
I've filed an issue for it: OPENDNSSEC-838 so that we can look into it.

With kind regards,
Berry van Halderen

> Regards,
> 
> - Håvard
> _______________________________________________
> Opendnssec-user mailing list
> Opendnssec-user at lists.opendnssec.org
> https://lists.opendnssec.org/mailman/listinfo/opendnssec-user
>