[gentoo-amd64] Re: Mobo recommendations

[Duncan]

Lee Thompson posted
<20051026051805.13619.qmail@×××××××××××××××××××××××.com>, excerpted below,
 on Tue, 25 Oct 2005 22:18:05 -0700:

> I'm attempting a dual core upgrade on a MSI Neo2
> Platinum.  It requires a BIOS upgrade which just won't
> boot linux as far as I can tell.  Anyone got a Socket
> 939 Mobo which will run 64 bit gentoo with a dual
> core?
>
>     http://forum.msi.com.tw/index.php?topic=89036.0

MSI isn't all that Linux friendly.  When I was shopping for a dual Opteron
board, I checked out the MSI site and rejected them because all the BIOS
upgrades and etc were in MSWormOS executable format (probably self
extracting zip), as was all the documentation.  I mailed them asking if a
virus/worm had gotten to them and replaced all the downloadables with
malware executables or what? Where were the platform neutral standard PDF
docs and zipped BIOS upgrades?  This was in fact before anything but beta
64-bit MSWormOS was available, but Linux already supported AMD64 standard,
so I found it a very bad omen that everything including the documentation
was proprietaryware MSWormOS executable format.

It's therefore not surprising to me at all that certain MSI BIOS upgrades
would have issues with Linux.  They probably spend very little if any time
testing them on Linux.

Tyan, OTOH, has the expected platform neutral standard PDF documentation
and zip file BIOS images.  As well, they have decently active Linux
support including Linux FAQs for most of their boards, lm_sensors
configurations, (sometimes proprietary) Linux drivers for RAID and other
devices, etc.  Many of their boards are Red Hat and SuSE/Novell certified
and proudly carry the certification marks in the manuals.

While the BIOS ugrading instructions asked for a DOS disk, last I checked,
I mailed them with my satisfactory results using the Ripcord OEM version
of FreeDOS, and suggested that they could supply complete bootable FreeDOS
based BIOS upgrade floppy images, as some manufacturers already do. 
(ASUS if I recall correctly, but that's from unconfirmed memory.)

With the Linux certifications, Tyan DOES test their BIOS images with
Linux, and mentions in the update description any problems they've found. 
(The latest for my board, for instance, mentions that those using the
onboard SATA in BIOS-RAID configuration will need to reinstall after the
upgrade.  Apparently, the new RAID format is incompatible with the old
one.  It doesn't mention it but of course, those using Linux software
RAID, with the BIOS set to standard single-drive config SATA wouldn't have
issues, since it'd be the kernel handling the RAID.)

So... I'd suggest buying a brand other than MSI next time you upgrade,
unless they decide to get serious about Linux before then, but of course
that doesn't help with your current situation...

That said and after reading the forum thread you linked above, mentioning
that the PATA side seemed to load fine, have you tried booting to either 
a PATA hard drive or a LiveCD, compiled with the SATA drivers as modules,
and loading them after successful boot?  There's no mention of that in the
forum thread.

If that works, you could then setup using either a PATA based / partition,
loading SATA and mounting your SATA partitions from there, or at minimum,
a PATA based initrd/initramfs, using it to load the SATA drivers and
switch to your real root on the SATA drive.

If it doesn't work, there's a fairly good chance you could at least get
more troubleshooting information logged before the crash, and go from
there, posting results to LKML (or kernel bugzilla) and asking for help
there, if necessary.

Another suggestion would be to try different kernels.  Try the earliest rc
with dual-core support.  Try the latest kernel in both vanilla and
mainline.  Try the latest kernel.org vanilla rc.  If you haven't found one
that works by then, try early 2.6 pre-dual-core support kernels.  You
won't have dual-core, but you might find one that works with the SATA on
the new BIOS, and can then try isolating the version in which it stopped
working and file an appropriate bug.

Also, if the chipset is supported anyway, try the old non-libata IDE based
kernel drivers.

-- 
Duncan - List replies preferred.   No HTML msgs.
"Every nonfree program has a lord, a master --
and if you use the program, he is your master."  Richard Stallman in
http://www.linuxdevcenter.com/pub/a/linux/2004/12/22/rms_interview.html


-- 
gentoo-amd64@g.o mailing list