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

Re: [Testbed-admins] tips device mapping

Mike Hibler wrote:

Several questions:

What version of FreeBSD are you running on the tipserv?


We're running BSD 6.3


Is there something running on the other end of the cable?  I.e., do you
have a "getty" enabled on the PC that the wire is connected to?  Or do you
have BIOS console redirect set and are watching the PC reboot?
We have our switches connected. Boss and ops are connected with a getty. The experimental nodes, which are the ones we're nominally trying to find, have BIOS console redirect. 


Are you sure the hood+wire are correct?
I'm not sure what you mean by this. We see the proper messages in dmesg. I've tried adding the ports to /dev/remote as you mention below, 
and then connecting with tip.  We see the same behavior as with cu.  We
get a "connected" message, but then no other interaction.  I'm a couple
hours drive away from the actual machines, so I can't really pull wires
out to narrow things down.

-Jim





The fact that the special files are there in /dev is a good sign that the
cards were found.  But look at /var/run/dmesg.boot and verify that there is
something like:

rp0: <RocketPort PCI> port 0xd800-0xd8ff,0xdc80-0xdcff mem 0xff7ffc00-0xff7ffc7f irq 17 at device 7.0 on pci2
RocketPort0 (Version 3.02) 32 ports.
rp1: <RocketPort PCI> port 0xd400-0xd4ff,0xdc00-0xdc7f mem 0xff7ff800-0xff7ff87f irq 13 at device 8.0 on pci2
RocketPort1 (Version 3.02) 32 ports.
rp2: <RocketPort PCI> port 0xcc00-0xccff,0xd080-0xd0ff mem 0xff7ff400-0xff7ff47f irq 16 at device 9.0 on pci2
RocketPort2 (Version 3.02) 32 ports.
rp3: <RocketPort PCI> port 0xc800-0xc8ff,0xd000-0xd07f mem 0xff7ff000-0xff7ff07f irq 2 at device 10.0 on pci2
RocketPort3 (Version 3.02) 32 ports.

I would start by unhooking all but one of the cables that interconnect the
PCI card(s) with the distribution panels.  That narrows the number of ports
you have to try.  Ours all appear as "/dev/cuaR<n>", but that is with an
embarassingly old version of FreeBSD (4.11).  Anyway, if you connect to
/dev/cuaR<whatever>, that should correspond to either the first (1) or last
(16 or 32) labeled port on the distribution panel.

"tip" is the slightly less ancient way to talk to serial ports.  If you put
something like:

	port0:dv=/dev/cuaR0:br#115200:pa=none:

in /etc/remote and then do:

	sudo tip port0

or, to try at 9600 baud:

	sudo tip -9600 port0

On Wed, Jun 17, 2009 at 03:40:18PM -0400, Hajime (Jim) Inoue wrote:

I'm trying to set up a tipserv remotely for a ~90 node emulab.
tips is set up with a 128 port rocketport.  I've rebuilt the
kernel with the proper device drivers.  The devices show up as
cuaRPI* and ttyRPI*, with the tty's having init and lock devices
as well.

My problem is that I do not know how to debug the setup properly,
and how the physical ports map onto the devices.  We have several
nodes powered up, so I've been trying to use the cu command to
connect to those devices.  We've tried

cu -s 115200 -l ttyRPIx

where x is the port we hypothesize our booted node is connected
to, but we don't get a response.  We do get "connected".  But no
interaction is seen.

We've tried various combinations of speeds 9600 and the cua* devices
as well.
Can anyone suggest a way to determine if the device drivers are configured properly? And also, how to map which devices are 
connected to which port?

Thanks,
-Jim
_______________________________________________
Testbed-admins mailing list
Testbed-admins@flux.utah.edu
http://www.flux.utah.edu/mailman/listinfo/testbed-admins