You need to watch dmesg, and as mentioned you might even try to keep the transfer enabled in the background with the command given a few posts before.
Any error you get in DMESG should be handled as Raspberry PI Bug and filed in their corresponding bugtracking system or even the Linux Kernel Mailinglist.
When asking us for support please prepare all information you can collect, we are even willing to forward the information to the corresponding people.
The tvheadend or driver logfile is not enough, yes it shows that the datastream is gone, but it does not reveal what is happening on your system, at least dmesg is also needed.
Currently our guess is that the specific USB Host controller triggers an interrupt at the wrong time and the host controller implementation runs against the wall. NULL pointer issues in the kernel are usually avoidable and software bugs from the Linux Kernel system (not even from our driver or application).