[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: serial console



I experimented some more.

I got the western scientific box to work at 115.2K, so yes it works.
It's just as hard to get into the bios setup screen; still have to
press <del> just before the first text appears.  The screen refresh
does work a bit different: instead of scrolling constantly it refreshes
once a second.  The effect is mainly that just over double the data is
output, and presumably it's a slightly more faithful copy of what's in
the vga video memory.  This does make it slightly more responsive to
key presses, but I can't say that this actually makes the bios all that
much more useful.  Actually, the most useful advantage I found is that
linux itself is slightly more fun at 115.2K.  The disadvantages are
that this has to be configured at least 3 different places (bios, lilo,
getty) to take effect, and it would considerably complicate our
operational environment in other ways.  For instance, all our boot cds
come up at 9600 baud.  I experimented a bit with a remote 9600 baud
connection connecting to the 115.2K connection to the western
scientific.  I didn't take much care with flow control, and this was
bad: the resulting connection dropped so many character it wasn't
useful and actually left the machine hung somehow.  Somebody with a
good 28.8K connection and/or working flow control *might* be able to
deal with a remote 115.2K console (at least until linux boots up), but
with a slower modem or a bad phone connection I'm doubtful the results
will prove useful.

In case anyone else wants to look at this, I've left
crossroads.rsug.itd.umich.edu setup at 115.2K baud + quick
boot=enabled.

rpaditya@xxxxxxxxx mentioned linuxbios.  Yup, that's interesting, and
there's at least 2 other projects that are also interesting:  openbios,
which aims to produce an opensource IEEE 1275-1994 bios, and there is
also FreeBIOS, which is merely trying for basic bios functionality.

			-Marcus Watts
			UM ITCS umce