[3093] in linux-net channel archive
Re: /dev/cua? Vs /dev/ttyS?
daemon@ATHENA.MIT.EDU (Matthias Urlichs)
Mon Jun 3 09:46:42 1996
From: Matthias Urlichs <smurf@smurf.noris.de>
To: linux-net@vger.rutgers.edu
Date: Mon, 3 Jun 1996 14:16:52 +0100
In linux.dev.net, article <199605141447.KAA14364@netaxis.netaxis.qc.ca>,
Mardy Hutchinson <mardy.hutchinson@netaxis.qc.ca> writes:
>
> There is one other reason to keep the cua devices: If you want to use
> setserial to change the parameters of a port that is open (usually with an
> idle getty on it...), then the cua device is available, when the ttyS device
> is not.
mgetty (in fact, any reasonable getty program) opens the port in non-
blocking mode, therefore /dev/cuaX is unavailable. Mingetty doesn't need
any modem control in the first place. Using any other getty, particularly
one that opens the port in blocking mode, is a bad idea, IMHO. (System
hangs (happens to the best...), customer calls, modem answers, customer
can't talk to the computer, customer gets angry, customer changes
provider, and Linux gets a bad name.)
I've deleted /dev/cu* on our systems. What's more, I'm for removing support
for /dev/cu* (or making it optional) in 2.1.
--
Virtue itself often offends when coupled with bad manners.
-- Middleton
--
Matthias Urlichs
