[Beowulf] A careful exploit?

Robert G. Brown rgb at phy.duke.edu
Fri Jun 14 05:06:31 PDT 2019


On Fri, 14 Jun 2019, John Hearns via Beowulf wrote:

> Regarding serial ports - if you have IPMI then of course you have a virtual
> serial port.
> 
> I learned something new about serial ports and IPMI Serial Over LAN
> recently..
> First of all you have to use the kernel config option ? ? console=ttyy0
> console=ttyS1,115200
> This is well known.

But, don't you have to have a LAN first?  A bit of a catch 22 if you're
trying to debug a LAN...;-)

> In the bad old days you had to edit the /etc/inittab and arrange a spawn of
> a getty process on /dev/ttyS1
> One gotcha - /dev/ttyS1 usualy corresponds to serial port 0 in the BIOS. (I
> may have that work=g but there is a mismatch)
> 
> These days there is a service which is managed under system and will
> automatically detect and configure the serial terminal.
> You still need the kernel console= option of course.
> system enable serial-getty
> 
> The youth of today etc. Not having to solder up RS232 plugs and find the
> baud rate by listening to a modem...

Solder?  Sure you mean crimp onto those itty bitty pins and sockets
before poking them through holes.in a half-shell, then not bothering to
put back the base... and where a lot of a time, you only really needed
the data lines (a good thing because you only HAD the data lines in
twisted pair, plus the ground sheath that yeah, you connected across to
the shell.

There are some sorta-amusing stories I recall from those days, such as
when somebody used this arrangement on two systems that were on two
circuits with a ground loop voltage of a few volts in between them due
to miswiring of a chassis.  I never did this myself, but I heard it was
fourth of july-ish.

I've heard that one can use a USB port as a serial com port, but have
never tried it, largely because I stopped doing clusters about the time
they dropped the real serial port and I never needed to.  But they were
incredibly useful, to the point where I had a serial port terminal
server in my cluster -- you could telnet/ssh into it and hook into a
serial console on any of the cluster nodes, or put your one serial
terminal on it and switch to any node while sitting in the server room.
I guess things are more automated and boilerplate now, but back then one
usually installed a stock linux and then provisioned it to meet your
needs and usually had to do the end stage configuration by hand.  It was
serial terminals or moving a video plug from system to system (although
I did plenty of that too, had a switch box for that too).

    rgb

> 
> Blog here on serial consoles in system
> ?http://0pointer.de/blog/projects/serial-console.html
> I am genuinely impressed at how good this is - it worked first time on? Debian
> system
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> On Fri, 14 Jun 2019 at 02:39, Robert G. Brown <rgb at phy.duke.edu> wrote:
>       On Thu, 13 Jun 2019, Jonathan Engwall wrote:
>
>       > I did not see Robert Brown's reply until J?rg Sa?manshausen
>       quote the entire
>       > thing.It's all gone now. I got rid of the netgear router too.
>       My policies
>       > are Drop and my rsa keys are specific. It works perfectly when
>       nobody slips
>       > in.
>       > "Hang a console on it" this sound fscinationg. Consoles give
>       me trouble.
>       > I will read up on it.
>
>       In the old days -- and I'm a relic of the old days;-) -- one did
>       EVERYTHING from a console.? Even now I do most of my systems work
>       from a
>       tty interface, basically an xterm.
>
>       It's a lot harder to manage now, since serial ports are dead...
>       if the
>       particular problem you're trying to debug is the network...
>
>       ? ? rgb
>
>       > Thank you.
>       > Jonathan Engwall.
>       >
>       >
>
>       Robert G. Brown? ? ? ? ? ? ? ? ? ? ? ? http://www.phy.duke.edu/~rgb/
>       Duke University Dept. of Physics, Box 90305
>       Durham, N.C. 27708-0305
>       Phone: 1-919-660-2567? Fax: 919-660-2525? ? ?email:rgb at phy.duke.edu
> 
>
>       _______________________________________________
>       Beowulf mailing list, Beowulf at beowulf.org sponsored by Penguin
>       Computing
>       To change your subscription (digest mode or unsubscribe) visit
>       https://beowulf.org/cgi-bin/mailman/listinfo/beowulf
> 
> 
>

Robert G. Brown	                       http://www.phy.duke.edu/~rgb/
Duke University Dept. of Physics, Box 90305
Durham, N.C. 27708-0305
Phone: 1-919-660-2567  Fax: 919-660-2525     email:rgb at phy.duke.edu




More information about the Beowulf mailing list