Summary from the tty side: - tty changes as proposed are an ugly hack - I'm opposed to magically producing device nodes that were not there before - the pty driver doesn't use ->driver_data which is probably what you want for some of the back walking - tty and pty code in this area is currently in flux - in the -next tree take a look at tty_kref_get and friends, passing kref bumped objects back may also help get a tidier result I'm currently trying to extract out all the pty nasty special case crap from tty_io.c which should actually help get this right. I'd also rather see big changes to the tty open/close logic where neccessary in favour of further hacks on hacks. Alan _______________________________________________ Containers mailing list Containers@xxxxxxxxxxxxxxxxxxxxxxxxxx https://lists.linux-foundation.org/mailman/listinfo/containers