Michael.Young@Cmu-10a (10/09/82)
From: Michael Wayne Young <Michael.Young@Cmu-10a> Date: 8 October 1982 1334-EDT (Friday) Recently, I ran into several such processes running around for my login... the problem I had was that a write was waiting to complete on the tty. When the line re-connected, the output buffer flushed, and the processes actually died (and init restarted a getty). You might want to have tty device drivers flush their output buffers if they detect anomolous conditions (such as a signal change on the RS232 line, in one of the various places), or if they haven't completed for a long time. [It is assumed that ANY write should complete after a given period of time -- anyone who ^S's something is simply holding things up -- people often use that to keep a line connected.] Michael