termios/PTY problem

[Till Straumann]

Fernando RUIZ CASAS wrote:

>Hi.
>You can do it add a task delete callback funtion.
>
While this would be feasible, it makes porting existing
code / writing portable code a nuisance.

BTW: I believe it's really a termios driver issue rather
than a pty one. If termios provided the hooks, it would
be easy for pty to use them. (As I said, I'm currently using
a hack)

Till

>
>Almost this is my opinion.
>The PTY is only a driver low_level. 
>A very reduced implentation to give at rtems the telnetd posibility.
>
>If you need a new funcion when EOF it's possible adding it.
>
>BRGDS
>
>Fernando RUIZ CASAS
>home: correo at fernando-ruiz.com
>work: fernandoruiz at wanadoo.fr
>
>
>>-----Mensaje original-----
>>De: Till Straumann [mailto:strauman at SLAC.Stanford.EDU]
>>Enviado el: viernes, 06 de septiembre de 2002 22:06
>>Para: rtems-users at oarcorp.com
>>Asunto: termios/PTY problem
>>
>>
>>Hi.
>>
>>I don't like the way the PTY driver reacts to the
>>network connection being closed (from the remote
>>side).
>>Currently, it just closes stdin&friends and deletes
>>the task that is using the PTY. This gives that
>>task no chance to cleanup resources.
>>
>>IMO, it should rather propagate an EOF condition
>>upstream.
>>
>>Unfortunately, however, the termios driver seems
>>not to be capable to convey EOF detected by
>>a lowlevel driver to the caller of rtems_termios_read()
>>and rtems_termios_write() etc.
>>
>>Shouldn't we fix this? I have a patch for pty to hack
>>around this (it also does NAWS and provides TIOCGWINSZ)
>>but IMHO it's really a termios problem.
>>
>>Till.
>>
>>
>>