Subject: Re: 2.0 on DEC3000 drops to debugger if console is not connected (resend)
To: Michael Smith <smithm@netapps.com.au>
From: Greg A. Woods <woods@weird.com>
List: port-alpha
Date: 01/16/2005 16:01:12
[ On Friday, January 14, 2005 at 21:21:36 (+0000), Michael Smith wrote: ]
> Subject: Re: 2.0 on DEC3000 drops to debugger if console is not connected (resend)
>
> On Tue, 4 Jan 2005 18:15:59 -0500 (EST)
> "Greg A. Woods" <woods@weird.com> wrote:
> > 
> > You might also want to build a kernel with
> > 	options DDB_FROMCONSOLE=0
> > so that BREAK is ignored by default.
> 
> I finally got my hands on the source code and made this change. I can
> now start my system without the console connected.

That's great to hear!

Now, can you do "sysctl -w ddb.fromconsole=1" after boot and then both
be able to drop the system to DDB on demand with a manually generated
BREAK signal?  If so then I'd think your hardware is OK (enough) and
that the unexpected BREAKs were manifestations of electrical noise, one
way or another.  (The RS232 line driver chip(s) or their related
discrete components may still be marginal, but still functional enough
to work correctly under "ideal" circumstances.)

Note that I would fully expect inserting and removing the console cable
to still be the cause of random unexpected BREAK-like signals, so if
you're going to disconnect the cable then I'd turn ddb.fromconsole off
again first.

That presents the dilema though that you'll be unable to force the
system into DDB if you don't leave the console cable connected and leave
ddb.fromconsole turned on.  This is why I try to have permanent console
connections to a terminal server.  :-)

-- 
						Greg A. Woods

H:+1 416 218-0098  W:+1 416 489-5852 x122  VE3TCP  RoboHack <woods@robohack.ca>
Planix, Inc. <woods@planix.com>          Secrets of the Weird <woods@weird.com>