[conspire] Sun console BREAK, and ... (was: Re: EIA RS-232-C, ... Re: DE-9, not DB-9 (was: ...

[Michael Paoli]

> From: Texx <texxgadget at gmail.com>
> Subject: Re: [conspire] EIA RS-232-C, ... Re: DE-9, not DB-9 (was: ...
> Date: Fri, 8 Mar 2019 19:24:50 -0800

> I remember that oddball serial port sun used.
> It was basically a way to avoid installing a second serial port on the
> chassis itself by hanging the other serial port off the secondary channel
> and THAT ONE DID have equal speed between the 2 channels.
> There was a problem with it though.
> If you had a laptop connected to it and you rebooted the laptop, it would
> do something oddball to the handshake pins and cause the server to halt.

Uhm, well, more generally, it was an operational hazard with Sun
serial consoles.  You see, with serial console on Sun, if you send
a BREAK signal, that's an out-of-band signal on the serial line,
if I recall correctly, it's >=300ms of solid mark - or space (signal
high - or low - I forget which - one of the two).  The BREAK signal
was relatively important with RS-232-C serial communication - most notably
because it would reliably work regardless of the baud rate.
And, commonly on Unix terminals, it was used to tell the host,
"Uhm, ... try your next configured baud rate" - typically to get the
host to try different baud rates - until hopefully getting one matching
the connected terminal.
Anyway, Sun Microsystems and serial console port - not an issue for other
serial ports, but just serial console - get a BREAK there, that tells
the operating system - at very low level - to halt - it would drop
to an
ok>
hardware monitor prompt, having halted the running operating system.
Great feature when you need it.  Inconvenient misfeature if that
unintentionally happens.  Well, anyway, some equipment, when connected
to serial port, when powered up/down or power cycled, what would come
across the serial port would look like a BREAK signal ... oops.
So, for example, some serial terminal servers, would specifically specify
and advertise that they were "Sun safe" - or the like - basically you could
power 'em up/down or pretty arbitrarily muck with their power, and they
wouldn't put out on the serial ports something that would look like
a BREAK signal ... and halt the attached Sun systems that had their
serial ports going to these terminal servers.  ("Oops") ...
whereas with some makes and Sun serial consoles - all those hosts would halt!
Anyway, Sun did, sometime later, come out with a patch that allowed one
to change that behavior - so it could instead be some particular
character sequence, rather than BREAK, to invoke that behavior.
And, wee bit more on that ... the less Solaris experienced sysadmins,
when that would happen, they'd be like "oh crud" ... and would reboot
the host (or power-cycle it to do so).  The bit more experienced sysadmins
would simply type:
go
at that
ok>
prompt, and the operating system would continue from exactly where it had
left off before being halted.

Also, sometimes one would get that BREAK effect when mucking about with
Sun serial console cabling (connecting/disconnecting/...).
Essentially BREAK on the serial console was the equivalent of
Stop-A on the Sun graphics console keyboard.