[CRIU] Re: [PATCH 10/10] tty: Don't try to restore controlling terminal via ptmx

Cyrill Gorcunov gorcunov at openvz.org
Fri Oct 26 16:38:28 EDT 2012


On Sat, Oct 27, 2012 at 12:32:52AM +0400, Pavel Emelyanov wrote:
> On 10/26/2012 11:58 PM, Cyrill Gorcunov wrote:
> > 
> > The slave peer should be restoring the controlling terminal.
> > It works for most of programs but need some more investigation
> > in future.
> > 
> I don't understand this change. Elaborate please.

Imagine we have two peers, master and slave, and it's controlling
terminal. With current code there will be two attempts to setup
controoling terminal -- at first for master peer and then for slave
peer. Usually programs setup it via slave peer in such case (but in
real I need more soficticated check here, I need to find a session
leader in tree and find which peer belond to it). This is a fast
fix for most of programs but one more will be done on top.


More information about the CRIU mailing list