Tim Glaremin wrote:
Hope it helps.... The part Marek quoted at least did change :)
The biggest problem is, that it is relative unreadable due to the reformatting done by your mua. The next big problem is that the guessing will begin now using the information in the stacktrace.
My first guess would be hardif_min_mtu. Maybe the netdev we want to access was removed in the time we try to update our information of the removed/deactivated interface (or another interface we also received some kind of notification).
I would have guessed that this cannot happen because we haven't processed all notification items, but I am awake a little bit too long to make 100% correct statements.
Maybe because we use dev_put in hardif_deactivate_interface too early and iterate over the already 'putted' device in our list (will be removed in hardif_remove_interface from our list) - which seems to be wrong too me right now. So all references will be gone when the interface goes down (so maybe it will be completely deleted at that point), but the pointer to the destroyed netdev is still in our list and will only deleted when the device will really be unregistered.
thanks, Sven