[ENet-discuss] freeing bug causing crash in enet_protocol_remove_sent_reliable_command

David Frampton davidframpton at majicjungle.com
Sat Jul 13 15:57:38 PDT 2013


Just following up on this. I wasn't able to get any more useful diagnostic information out of it as it was such a rare issue.

But I think I may have found the cause. It turns out I was occasionally calling enet_peer_send from a different thread. I've now changed it to send that data on the main thread and haven't seen any problems so far.

To answer your question, I am using 1.3.7.

Cheers,
David

On 11/06/2013, at 6:39 PM, ketmar <ketmar at ketmar.no-ip.org> wrote:

> On Tue, 11 Jun 2013 17:28:28 +1200
> David Frampton <davidframpton at majicjungle.com> wrote:
> 
> it looks strange. i can suggest to rebuild enet with -O0 -g and try to
> run your app with valgrind: it may trace the origins of this pointer.
> 
> note that valgrind can 'hide' crash conditions (i.e. app can go past
> the 'crash point'), so be careful and watch valgrind logs
> (--log-file=xxx)
> 
> btw: what version of enet are you using? is it stock 1.3.7 or 1.3.8?
> 
> also, make sure you aren't accidentally free/reuse either ENetPacket or
> it's 'data' field after passing it to enet_peer_send.
> _______________________________________________
> ENet-discuss mailing list
> ENet-discuss at cubik.org
> http://lists.cubik.org/mailman/listinfo/enet-discuss



More information about the ENet-discuss mailing list