mgetty on ptys? (On SCO 5.0)

• Messages sorted by:[ date ][ thread ][ subject ][ author ]

Hello,

Robert Barrell wrote:
>      Currently, I am trying to get mgetty version 0.98 to run under SCO ODT
> Release 5.  It works okay on a regular serial line modem port, but the modems
> for which we would like to implement Caller-ID are attached to Annex-XL
> terminal servers using rtelnet to establish connections as pty's.  

Interesting. Didn't know that was possible :)

> The problems we are encountering are two:
> 
>  1) On serial lines or rtelnet lines alike, the first call after mgetty is
>     brought up on the port is rejected as being a bad Caller-ID (even when
>     the Caller-ID is valid).  On serial lines, this happens every other call:
>     first, third, etc.).

Hmmm. Unfortunately, you didn't send me a mgetty log, so I can't see
whether the caller ID was never sent to the mgetty process at all, or just
not handled correctly (both is thinkable).

>From your description, this sounds somewhat like a modem or Telco problem.

>  2) On rtelnet lines only, after the second call is successful, mgetty never
>     fully clears the line to make room for the third call.  The logfile shows
>     that it gets as far as "waiting for line to clear (VTIME), read:...." but
>     it just sits there reading without timing out.  I have tried changing the
>     VTIME value, compiling with -DBROKEN_VTIME (in which case I get the same
>     thing, but with "(select)" instead of "(VTIME)"); and have even tried
>     using an alarm call (which does time out, but keeps on looping without
>     getting back to the "waiting..." state).  Most of the time, if I stop and
>     restart rtelnet for the pty to the modem, things will reset enough to
>     allow a new call (starting back with problem (1) again, though).

This may hint at a problem with the select() / VTIME implementation on
this rtelnet / pty implementation. I do not have enough experience with
this to say for sure, though.

Please try -DBROKEN_VTIME -DUSE_POLL, this will use poll() instead of
select(), maybe it will change something.

If it won't help, I'd suggest posting to comp.unix.sco.*, since (to my
knowledge) no system gurus using SCO 5.0 with mgetty are reading this
list [but I may err].

gert

-- 
                                                            //www.muc.de/~gert
Gert Doering - Munich, Germany                             gert@greenie.muc.de
fax: +49-89-3545980 <---new!!!              gert.doering@physik.tu-muenchen.de