"Arun K. Khan" wrote:
>
> Scroll down for my response.
>
> At 01:05 PM 12/18/98 CST, David Douthitt wrote:
> >> At 10:57 AM 12/18/98 +0000, you wrote:
> >> >I've just been experimenting with puting a getty dialin on the same
> >> >modem as diald uses to get out. I think I've probably done something
> >> >wrong (which I'll see if I can sort) but it has provoked what I reckon
> >> >must be a diald bug related to locking. The sequence is as follows:
> >
> >> If I understand the post correctly, the same device is being used
> >> for a dialin as well as dial out; both daemons are trying to
> >> lock the device to insure exclusive use of it. Select unique
> >> devices for diald and getty resp. each will need its dedicated
> >> modem device.
> >
> >Just isn't so!
> >
> >I am happily using mgetty, minicom, and diald (more than one!) on the
> >same port. However, the original sender did not detail which OS
> >they were using.
>
> The original poster, I believe was having problems with lock files. Diald
> daemon trying to lock the device and finding /var/lock/LCK..cua3 owned by
> another PID. From the device name, it appears the OS is Linux (FreeBSD
> serial device names are /dev/cuaa[0-3], cuaia[0-3], cuala[0-3]).
>
> Until now, I was under the impression, that once a specific device is
> locked it cannot be used by another process (to ensure availablity of
> service by the locking process). If the **same** device can be used by
> (each using it's locking mechanism) mgetty/diald/minicom respectively then
> I am interested in finding out how it can be done. I would appreciate if
> you could share your configuration info for the above applications. It
> would be another selling point for Linux.
>
> >Under Linux, the ports used should be /dev/ttyS* (and NOT /dev/cua*). Under
> >FreeBSD (so I have heard) /dev/cua* should be used.
>
> I have not seen any clear views on this. Some say /dev/cua* are the old
> serial devices provided for backward compatibility and that only /dev/ttyS*
> should be used to call out/in. Some say, use ttyS* to call in and cua* to
> call out. Fr FreeBSD serial devices see above.
Under Linux the cua devices are older and lack some of the functionality
of the ttyS devices. The cua devices are to be phased out soon.
Confusion
remains. The ttyS devices can be used for everything that I have seen.
So the suggestion is to use ttyS devices to configure all serial port
services so that locking can work. I have read this in several places.
If symlinks to serial devices are used there can be more confusion.
Several Linuxes have a symlink called /dev/modem. Locking /dev/modem
does not lock what it points to. So if one is going to use /dev/modem
in configuration files, you can't use its target in other configuration
files.
In any case, if /var/lock locking is to work, you have to decide which
set of devices to use, and use it consistently across all services
that use serial ports. So I would say it is not a diald bug, its a
configuration error.
>
> Happy Holidays,
> Arun Khan
>
> -
> To unsubscribe from this list: send the line "unsubscribe linux-diald" in
> the body of a message to [EMAIL PROTECTED]
--
Jan Carlson
[EMAIL PROTECTED] Scarborough, Ontario, Canada
Mailed with Netscape 4.5 on Red Hat Linux 5.2
-
To unsubscribe from this list: send the line "unsubscribe linux-diald" in
the body of a message to [EMAIL PROTECTED]
