Hi,
I don't have that motherboard, but I do have a test machine with the EM64T.

First off ...  there isn't a smp kernel for that chipset.  That doesn't mean
it doesn't support smp, that just means that for that kernel, they didn't make
the kernel say 'smp'.  It's the same for Itanium cpu's as well.
So if you are booting into a smp kernel for that ... then it's the smp kernel
for an Opteron system, so things are going to be a little different and
possibly fail.

So ... how to check that you've got the right kernel installed.
You can do a rpm command that will list the arches, or, in my opinion a much
easier way is to do a

   yum list "*kernel*"

And look and see if your kernel is a ia32e or a x86_64.  It should be a ia32e.

Second off.  Was this a fresh SL 303 install, or an upgrade from SL 302?
On SL 302, I had a similar problem you had, but it was because the installer
would load the ia32e regular kernel, and then the x86_64 smp kernel, and the
installer would tell the machine to boot up off the smp kernel.
I haven't had this problem when doing a fresh SL 303 install, so I'm wondering
if that got completely fixed, or just fixed on some motherboards.

Troy

Michael David Joy wrote:
> Hello All,
> I just finished installing SL 3.03 X86_64 on a Supermicro X6DAL-G board
> (Intel 7525 chipset). This is a dual Xeon EM64T 2.8 Gigahertz system. I
> noticed two strange problems. One is with the Silicon Image 3112 based
> controllers. I could not get any drives to format more than about 2-3
> percent on 3 different manufacture PCI SI3112 serial ata controllers.
> The drives formatted just fine with the Seagate sata tools in dos on
> this controller, but not in linux. The same drives format fine with the
> ATA_PIIX driver using the on board ICH5 sata controller in linux.
>
> This seems to be an problem related to this particular intel chipset and
> SI3112 driver as it happens with Fedora Core 2 x86_64 as well as FC3
> Test 1 x86_64.
>
> On a separate note, if I boot this box with the included 2.4.20 SMP
> kernel in SL3.03 the system fails to initialize the ATA_PIIX controller,
> kernel dumps some error messages and immediately reboots. If I boot with
> the non-smp kernel, the system comes up just fine, and show's 4 cpu's as
> you'd expect from a hyperthreaded dual cpu system. However, why does the
> non-smp kernel show dual processors, and why does the SMP kernel hang?
>
> I know there's probably not an easy answer to this question. Any ideas?
>
> --
> Michael Joy
> HEP - University of Mississippi
> [log in to unmask]


--
__________________________________________________
Troy Dawson  [log in to unmask]  (630)840-6468
Fermilab  ComputingDivision/CSS  CSI Group
__________________________________________________