[CRIU] [PATCH 1/5] tty: Introduce TTY_TYPE

[Cyrill Gorcunov]

On Wed, Oct 15, 2014 at 02:02:54PM +0400, Pavel Emelyanov wrote:
> On 10/15/2014 02:01 PM, Cyrill Gorcunov wrote:
> > On Wed, Oct 15, 2014 at 01:57:45PM +0400, Pavel Emelyanov wrote:
> >>>>>  	switch (major) {
> >>>>> -	case UNIX98_PTY_SLAVE_MAJOR:
> >>>>>  	case TTYAUX_MAJOR:
> >>>>> -		return tty_types[major];
> >>>>> +		if (minor == 0 || minor == 2)
> >>>>
> >>>> The comparison of minor first appeared in this patch which makes
> >>>> it not just "introduce the type" but also "fix some bug".
> >>>
> >>> Nope. Not testing minors earlier simply defering exit, 
> >>
> >> Where is this place currently in the code?
> > 
> > parse_pty_index, /dev/console get treated as master pty peer and
> > of course ioctl doesn't work on it (because /dev/console is a special
> > slave peer)
> 
> That's minor == 1. How about the other minors?

Well, we didn't meet other minors previously so I don't know how
criu would behave there, if the terminal is pty then it'll pass,
if no -- fail. Here we require explicit minors match to make
more predictable behaviour.