a fun project

[Don Wilde]

On Sat, Mar 29, 2008 at 7:17 AM, JoaoBR <joao at matik.com.br> wrote:

> On Saturday 29 March 2008 06:05:49 Don Wilde wrote:
> > FreeBSD working on such bleeding edge hardware leads to more high-end
> sales
> > for Dell and more FreeBSD in High Places. Win - Win!
>
>
> what a generous offer!
> :)
>
> --
>
> João
>
> Okay, now it's time for help lifting the cart out of the mudhole! :)

Mixed results on the first pass. The blade is mostly ACPI driven, with lots
of PCI bridges and PCIexpress mezzanine card slots. Probing found the mpt0
LSILogic SAS6 controller, so I should be able to at least load the hard
disk.

I was able to boot the distro kernel of  amd64-7.0-R from a USB CDROM, but
it hung up three quarters of the way through the probe process. It hung
after the last PCI probe, and before the first SIO probe. The 'verbose
logging' bootup didn't give me any more clues; it identified the 34th (!)
PCI bus port more completely, but it didn't give any more clues to what hung
it up.

I was able to get the i386 ISO to complete the probe and move on to the
FreeBSD sysinstall menu, but wasn't able to complete the install due to the
fact that there are no ATAPI CDROM ports on the blade and it couldn't use
the USB CD as a source.

Any ideas as to why the amd64 probe hung and the i386 did not? My i386 ISO
set is actually 7.0-RC3, not 7.0-RELEASE, would that matter?


I have two ways to proceed in loading the system from the i386 set. Our
blades have what we call vMedia, which allows us to virtualize a CDROM or
floppy drive and attach it through our iDRAC blade management processor.
We've got a few bugs left in that part of this new blade's firmware, so that
may not be successful although it would be the easiest way to succeed if it
would work.

The other way to proceed is with a local ftp server. Our lab networks (and
corporate, for that matter) are heavily blockaded by Doze-oriented
sysadmins, and we have very little outside access capability. What I think I
can do, however, is to preload my trusty little Inspiron laptop as a partial
clone of ftp.freebsd.org, and attach it to the local net segment when I go
in tomorrow. That way, I think I can at least get the system loaded with a
running i386 setup so that I can capture probe dmesgs and run some test
scripts.

I feel good about what I saw today. I'm thinking that it can and will be
made to work successfully, because Dell's  hardware is very much
standards-driven and should be straightforward even if it is all bleeding
edge for the PC world.

Comments or suggestions? I suppose I could d/l an i386 live CD and boot that
as an intermediate step. At least with that I could get quickly to a prompt
and mount a USB key to store off dmesg and stuff like that that I can't
capture with the incomplete install boot.

-- :D