[CRIU] Re: [PATCH 2/2] tty: Add checkpoint/restore for ttys

[Cyrill Gorcunov]

On Mon, Jun 18, 2012 at 10:56:51AM +0400, Stanislav Kinsbursky wrote:
> 18.06.2012 09:50, Andrey Wagin написал:
> >>>Can a child open a slave point by himself?
> >>Yes, child can open but it need to know when this peer is available for
> >>opening, so since the master process knows exactly when child peer is
> >>created, it opens it and send fd to a child process which waits for fd
> >>to arrive.
> >>
> >Ok. I agree with that.
> 
> Just my 5 cents.
> FD bouncing between processes is obfuscating by design.
> There are some situations, when we just can't do it otherwise. But
> sometimes it looks like overkill (I'm not contend about current
> situation since I haven't dig deep enough into it).
> And I believe, that we have to try to simplify CRIU logic as much as
> possible.
> CRIU is complicated enough already.

Well, I must admit it's not as simple as I would like it to be,
we could try to simplify it though.

	Cyrill