Tyan serial console - how?
mathog at mendel.bio.caltech.edu
Thu Nov 7 15:22:08 EST 2002
> Ken Chase wrote:
> I had massive problems playing with the serial console on linux at
> anything higher than 9600. Having no flow control is good, but more
> and more dangerous at higher bitrates.
Can anybody explain why there no flow control is good? The
cables in question have all pins, so naively, I would have
thought that hardware flow control would have been the best choice.
> Remember the screen is being repainted raster style from top to bottom
> in a sweep - like a tv's electron gun. At 115Kbps it might look like
> 'snow' as the errors appear for certain letters here and there and
> are corrected on the next sweep.
I've tried a few speeds. At 115Kbps the standard
BIOS displays are more or less faithfully rendered. At lower
speeds only chunks of them tend to be seen, unless they
linger on screen for several seconds.
This BIOS serial console mode really is a dreadful hack. Start up
memtest86, for instance, and nothing comes out the serial line,
it only shows up on the video. During a normal boot (if there
is such a thing on the 2466 mobo) you may or may not
see the bits of text you're looking for as the screen rewrites.
Phoenix/Award should come up with a real serial line BIOS (or
even a USB interface, since serial is now "legacy") which
can be used instead of the standard VGA BIOS, rather than layered
over it. Naively you'd think that the serial mode BIOS
would be both smaller and simpler than the current versions.
Sure, it isn't a huge market. But maybe the next time one
of the readers of this list configures a 10K node system
for some government lab a way can be found to pay for the
development of a real serial mode Bios?
> Go back to 9600 and try again.
Tried that - it was too slow in the BIOS part. Worked ok
in the linux serial console mode.
mathog at caltech.edu
Manager, Sequence Analysis Facility, Biology Division, Caltech
Beowulf mailing list, Beowulf at beowulf.org
To change your subscription (digest mode or unsubscribe) visit http://www.beowulf.org/mailman/listinfo/beowulf
More information about the Beowulf