Subject: Re: ptyfs fully working now...
To: Christos Zoulas <>
From: Michael Graff <>
List: tech-kern
Date: 11/17/2004 12:58:37
Is there any lsof-related issues with (say) finding out who is talking with a 
specific pty?  Or is that really not an issue?


On Thursday 11 November 2004 15:05, Christos Zoulas wrote:
> In article <>,
> Thor Lancelot Simon <> wrote:
> >On Thu, Nov 11, 2004 at 04:51:29PM -0500, Christos Zoulas wrote:
> >> Sure, that will work the same as before. The problem with such
> >> scripts is that what happens when someone is using ptyqf? I'd rather
> >> have a program that opens a pty dynamically and then makes a symbolic
> >> link from it to a known name. When the program exits, it can remove
> >> the symbolic link.  It is trivial to change the code to take the
> >> symbolic link destination name instead of the actual pty, and then
> >> use openpty() to allocate a pty dynamically.
> >
> >So, all ptys will always have names in the filesystem?
> The open slave sides should (it would be strange if they did not).
> The master side does not need to have one (and it does not).
> To answer the question differently. If you don't need the functionality
> of being able to open a particular pty by name, you can remove
> /dev/pty?? and /dev/tty?? [ahem, not literally because /dev/ttyE0
> is not a pty, but you get my drift].
> christos