Re: n_tty: Check the other end of pty pair before returning EAGAIN on a read()

From: Marc Aurele La France
Date: Thu Jan 14 2016 - 16:50:29 EST

On Fri, 18 Dec 2015, Marc Aurele La France wrote:
On Fri, 18 Dec 2015, Peter Hurley wrote:
On 12/18/2015 06:26 AM, Marc Aurele La France wrote:
On Fri, 11 Dec 2015, Peter Hurley wrote:
On 12/11/2015 05:37 AM, Marc Aurele La France wrote:

I am not asking to read data before it has been produced. I am puzzled
that despite knowing that the data exists, I can now be lied to when I
try to retrieve it, when I wasn't before. We are talking about what is
essentially a two-way pipe, not some network or serial connection with
transmission delays userland has long experience in dealing with.

These previously internal additional delays, that are now exposed to
userland, are simply an implementation detail that userland did not,
and should not, need to worry about.

Your mental model is that pseudo-terminals are a synchronous pipe, which
is not true.

But this argument is pointless because the regression needs to be fixed
regardless of the merits.

Fair enough.

Anything new on this?

It's on my todo list.

While considering this issue further, I was curious what ssh does
regarding the entire foreground process group and its output?

If ssh only knows that the child has terminated, how does it wait
for the rest of the foreground process group's output since those
processes may not yet have received their SIGHUP/SIGCONT signals
yet?

sshd cannot know about the termination of any process other than the
session leader because any of the session leader's children are
re-parented to init. The idea is to, at minimum, collect any output the
session leader might have left behind. Yes, this could entail also
collecting output from its children that might have squeaked in, but
that's gravy that can't be avoided.

This situation is much simpler on the *BSDs. There, both ends of the
pty pair are, in effect, completely closed after disassociation of either
end, preventing (with EIO) any further output (but still allowing data
already collected to be read, after which an EIO occurs). It's
unfortunate System V variants don't do this, but that's crying over spilt
milk.

Anything more on this?

Thanks.

Marc.