64 bit kernel

64 bit kernel

[Stan Gammons]

Hi everyone,

I was wondering if anyone can tell me if the following is a 64 bit
kernel problem or if it's a BIOS problem.

I have a Gigabyte K8NSC-939 with an AMD64 3200+ (Venice) CPU version F7
BIOS. When I first got this board, I put a single 512 Mb PC2700 DIMM in
it from an older Celeron board I had. 32 bit Suse 10.0 and 32 bit FC4
loaded fine. When I tried the 64 bit version of either, I kept getting
DMA errors on boot like the HD or controller was bad. After some
searching I found others with similar problems and they had to use
"noapic nolapic" kernel boot options to install and boot the OS. That
worked for me too and I was able to install the OS.

After I upgraded the memory and put 2 512Mb PC3200 DIMMS in the board. I
tried a 64 bit install again. This time I no longer had to use the
"noapic nolapic" options. With a single DIMM, BIOS (during boot)
reported "single channel" memory. With 2 DIMMS, BIOS (during boot)
reports "dual channel" memory. My question though is does the 64 bit
kernel require "dual channel" memory or is this a BIOS problem?  



Stan

Re: 64 bit kernel

[Mike McCarthy, W1NR]

I saw a similar issue many years ago that turned out to be a chipset bug. 
This was a PII system that used 16 bit wide modules.  When using only one 
module, the chipset "fooled" the OS into thinking that it was doing 32 bit 
wide operations.  However, it failed at full speed.  Reducing the memory bus 
speed or installing modules in pairs "fixed" the problem.  I suspect a bus 
or memory controller issue rather than the kernel.

The failure mode was exactly as you describe.  It manifested itself as disk 
errors or DMA failures.  Unfortunately the chipset vendor determined that it 
was a silicon bug and said that they would NOT fix it!

Mike

----- Original Message ----- 
From: "Stan Gammons" <s_gammons@charter.net>
To: "Linux Kernel Mailing List" <linux-kernel@vger.kernel.org>
Sent: Sunday, January 08, 2006 11:27 PM
Subject: 64 bit kernel


> Hi everyone,
>
> I was wondering if anyone can tell me if the following is a 64 bit
> kernel problem or if it's a BIOS problem.
>
> I have a Gigabyte K8NSC-939 with an AMD64 3200+ (Venice) CPU version F7
> BIOS. When I first got this board, I put a single 512 Mb PC2700 DIMM in
> it from an older Celeron board I had. 32 bit Suse 10.0 and 32 bit FC4
> loaded fine. When I tried the 64 bit version of either, I kept getting
> DMA errors on boot like the HD or controller was bad. After some
> searching I found others with similar problems and they had to use
> "noapic nolapic" kernel boot options to install and boot the OS. That
> worked for me too and I was able to install the OS.
>
> After I upgraded the memory and put 2 512Mb PC3200 DIMMS in the board. I
> tried a 64 bit install again. This time I no longer had to use the
> "noapic nolapic" options. With a single DIMM, BIOS (during boot)
> reported "single channel" memory. With 2 DIMMS, BIOS (during boot)
> reports "dual channel" memory. My question though is does the 64 bit
> kernel require "dual channel" memory or is this a BIOS problem?
>
>
>
> Stan
>
>
> -
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at  http://www.tux.org/lkml/
> 

Re: 64 bit kernel

[Gene Heskett]

On Monday 09 January 2006 08:25, Mike McCarthy, W1NR wrote:
>I saw a similar issue many years ago that turned out to be a chipset
> bug. This was a PII system that used 16 bit wide modules.  When using
> only one module, the chipset "fooled" the OS into thinking that it
> was doing 32 bit wide operations.  However, it failed at full speed. 
> Reducing the memory bus speed or installing modules in pairs "fixed"
> the problem.  I suspect a bus or memory controller issue rather than
> the kernel.
>
>The failure mode was exactly as you describe.  It manifested itself as
> disk errors or DMA failures.  Unfortunately the chipset vendor
> determined that it was a silicon bug and said that they would NOT fix
> it!
>
And that sucks the big one, Mike.  Will you share that vendors name so 
we can bypass them in future purchase thinking?

-- 
Cheers, Gene
People having trouble with vz bouncing email to me should add the word
'online' between the 'verizon', and the dot which bypasses vz's
stupid bounce rules.  I do use spamassassin too. :-)
Yahoo.com and AOL/TW attorneys please note, additions to the above
message by Gene Heskett are:
Copyright 2005 by Maurice Eugene Heskett, all rights reserved.

Re: 64 bit kernel

[Stan Gammons]

On Mon, 2006-01-09 at 08:25 -0500, Mike McCarthy, W1NR wrote:
> I saw a similar issue many years ago that turned out to be a chipset bug. 
> This was a PII system that used 16 bit wide modules.  When using only one 
> module, the chipset "fooled" the OS into thinking that it was doing 32 bit 
> wide operations.  However, it failed at full speed.  Reducing the memory bus 
> speed or installing modules in pairs "fixed" the problem.  I suspect a bus 
> or memory controller issue rather than the kernel.
> 
> The failure mode was exactly as you describe.  It manifested itself as disk 
> errors or DMA failures.  Unfortunately the chipset vendor determined that it 
> was a silicon bug and said that they would NOT fix it!

Hi Mike, 

What chipset was that?  

This board has an nVidia nForce 3 chipset on it.  This brings about
another question. What is the consensus on using the amd74xx.c patch for
the nForce 3/4 chipset that nVidia has on their website?  It's supposed
to improve HD performance. Any comments on including that patch in the
existing kernel?  How about the pros and or cons of adding that patch
and rebuilding a system specific kernel?



Stan

Re: 64 bit kernel

[Andi Kleen]

Stan Gammons <s_gammons@charter.net> writes:

First you get the price of the day for meaningless subjects. Gratulations.

> I was wondering if anyone can tell me if the following is a 64 bit
> kernel problem or if it's a BIOS problem.
> 
> I have a Gigabyte K8NSC-939 with an AMD64 3200+ (Venice) CPU version F7
> BIOS. When I first got this board, I put a single 512 Mb PC2700 DIMM in
> it from an older Celeron board I had. 32 bit Suse 10.0 and 32 bit FC4
> loaded fine. When I tried the 64 bit version of either, I kept getting
> DMA errors on boot like the HD or controller was bad. After some
> searching I found others with similar problems and they had to use
> "noapic nolapic" kernel boot options to install and boot the OS. That
> worked for me too and I was able to install the OS.

That's usually an ACPI problem. Put full boot log of the failure
and acpidmp output into bugzilla.kernel.org

The difference between 32bit and 64bit distro is that the later
use the APIC by default so they are more sensitive to ACPI issues.

> 
> After I upgraded the memory and put 2 512Mb PC3200 DIMMS in the board. I
> tried a 64 bit install again. This time I no longer had to use the
> "noapic nolapic" options. With a single DIMM, BIOS (during boot)
> reported "single channel" memory. With 2 DIMMS, BIOS (during boot)
> reports "dual channel" memory. My question though is does the 64 bit
> kernel require "dual channel" memory or is this a BIOS problem?  

That sounds weird. Most likely a BIOS problem of some sort.

Dual or single channel memory shouldn't make a difference to the kernel.
 
-Andi