🚀 go-pugleaf

Finally found the cause of a problem which I believe causes detailed 
trace logging to fail and also occasional crashes using
'dial sl0 dialer' at the net> prompt to redial after a line drop -
usually causing a complete hang and trashing log files.

Tested this with both net2.16d and my own 'lite' build; ps shows the 
sl0 monitor process running down to within 16 words free stack.

In 8250.c, asy_init(), line 159 of the 2.16d source:
(slightly reformatted for space)

  if (monitor)
	{
	/* set up to monitor line signals */
	ap->monitor = newproc (ifn = if_name (ifp, " monitor"),
		256, asy_monitor, ifp->dev, ifp, NULL, 0);
		^^^ change to 512
	free (ifn);
	pwait (NULL);	/* let hooks get set up */
	}

Increasing the stack size for the monitor process here fixes all the 
problems I was having - tried 1024 initially but having watched it, it 
only reached 363 words max on a full byte level trace running dialer 
twice, so I would suggest 512 as adequate.

I've now managed to successfully capture a packet & byte level trace 
of an entire session - I'd previously assumed it was down to not 
enough CPU power, despite having fairly fast drives ...

Cheers,

      Graham.
-- 
----[ Graham E. Kinns ]----==========----< Graham@grempc.demon.co.uk >----