Re: Asus A7M266-D, Linux 2.6.7 and APIC

Re: Asus A7M266-D, Linux 2.6.7 and APIC

[Mikael Pettersson]

On Tue, 20 Jul 2004 17:55:32 -0700 (PDT), Andy Biddle wrote:
>I have a dual-proc server that I've recently decided to overhaul.  It's an
>Asus A7M266-D motherboard.  It had been running with dual Athlon MP 1800+s
>and RedHat and BIOS rev 1003 for at least a year with no real problems.
>
>First I decided to change the OS to Gentoo.  I build the system with no
>problem and built a custom kernel based on linux 2.6.7-gentoo-r11.
>Everything was working great.  Then I changed out the procs and went with
>dual Athlon MP 2800+s.  To support these, I needed to (according to Asus's
>website) upgrade the BIOS to 1011.002 or higher.  (Latest is 1011.003, so
>that's what I used.)
>
>Now, when I boot to this custom kernel I get about 2 seconds into the boot
>process before the system starts spewing constant "APIC error on CPU0:
>04(04)" messages.   It stops booting at that point.
>
>I've done a little research and it seems that I'm supposed to do a couple
>of things:
>	1.  Disable "MPS 1.4 Support" in the BIOS.
>	2.  pass the kernel "noapic" as a parameter.
>
>I've done both of these and I STILL get the APIC messages.  (Which leads

04 is "send accept error". That noapic doesn't kill it means
that it occurs when CPU0 sends messages to CPU1.

I suspect that CPU1 isn't booting properly, or that the BIOS
set it up with an APIC ID not matching the MP table.

How far in the boot did it get before the errors & halt?

1. Try swapping CPU0 and CPU1.
2. Try downgrading both CPUs to your original MP 1800+ ones.
3. Check if ASUS' German ftp site has a newer beta BIOS than
   that 1011.003 you're using. For older boards these betas
   may be the only way to get newer CPUs to work.

Re: Asus A7M266-D, Linux 2.6.7 and APIC

[Andy Biddle]

Well I had a few minutes this morning, so I swapped CPU0 and 1 and still
have the same results.  Boots in single proc mode, APIC errors on a
dual-proc kernel.  (For what it's worth, both CPUs are recognized by the
BIOS and report "MP capable".

I checked the Asus website and I'm at the latest beta kernel...

Tonight I'll bounce back to the 1800+s to see if it's a BIOS thing...

Any other suggestions?

Thanks,
AndyB

On Wed, 21 Jul 2004, Mikael Pettersson wrote:

> On Tue, 20 Jul 2004 17:55:32 -0700 (PDT), Andy Biddle wrote:
> >I have a dual-proc server that I've recently decided to overhaul.  It's an
> >Asus A7M266-D motherboard.  It had been running with dual Athlon MP 1800+s
> >and RedHat and BIOS rev 1003 for at least a year with no real problems.
> >
> >First I decided to change the OS to Gentoo.  I build the system with no
> >problem and built a custom kernel based on linux 2.6.7-gentoo-r11.
> >Everything was working great.  Then I changed out the procs and went with
> >dual Athlon MP 2800+s.  To support these, I needed to (according to Asus's
> >website) upgrade the BIOS to 1011.002 or higher.  (Latest is 1011.003, so
> >that's what I used.)
> >
> >Now, when I boot to this custom kernel I get about 2 seconds into the boot
> >process before the system starts spewing constant "APIC error on CPU0:
> >04(04)" messages.   It stops booting at that point.
> >
> >I've done a little research and it seems that I'm supposed to do a couple
> >of things:
> >	1.  Disable "MPS 1.4 Support" in the BIOS.
> >	2.  pass the kernel "noapic" as a parameter.
> >
> >I've done both of these and I STILL get the APIC messages.  (Which leads
>
> 04 is "send accept error". That noapic doesn't kill it means
> that it occurs when CPU0 sends messages to CPU1.
>
> I suspect that CPU1 isn't booting properly, or that the BIOS
> set it up with an APIC ID not matching the MP table.
>
> How far in the boot did it get before the errors & halt?
>
> 1. Try swapping CPU0 and CPU1.
> 2. Try downgrading both CPUs to your original MP 1800+ ones.
> 3. Check if ASUS' German ftp site has a newer beta BIOS than
>    that 1011.003 you're using. For older boards these betas
>    may be the only way to get newer CPUs to work.
>
>

Re: Asus A7M266-D, Linux 2.6.7 and APIC

[Mikael Pettersson]

On Wed, 21 Jul 2004 08:46:11 -0700 (PDT), Andy Biddle wrote:
>Well I had a few minutes this morning, so I swapped CPU0 and 1 and still
>have the same results.  Boots in single proc mode, APIC errors on a
>dual-proc kernel.  (For what it's worth, both CPUs are recognized by the
>BIOS and report "MP capable".
>
>I checked the Asus website and I'm at the latest beta kernel...

No you're not. You have 1011.003, while the latest beta BIOS
is 1011 beta 05.

ftp.asuscom.de
pub/ASUSCOM/BIOS/Socket_A/AMD_Chipset/AMD_760_MPX/A7M266-D/

Re: Asus A7M266-D, Linux 2.6.7 and APIC

[Andy Biddle]

Mikael,

You were absolutely right.  There was an update that I couldn't see from
the website.  Even better news is that it totally solved my problem.  No
APIC problems at all now!

Thanks a TON for the help!


On Wed, 21 Jul 2004, Mikael Pettersson wrote:

> On Wed, 21 Jul 2004 08:46:11 -0700 (PDT), Andy Biddle wrote:
> >Well I had a few minutes this morning, so I swapped CPU0 and 1 and still
> >have the same results.  Boots in single proc mode, APIC errors on a
> >dual-proc kernel.  (For what it's worth, both CPUs are recognized by the
> >BIOS and report "MP capable".
> >
> >I checked the Asus website and I'm at the latest beta kernel...
>
> No you're not. You have 1011.003, while the latest beta BIOS
> is 1011 beta 05.
>
> ftp.asuscom.de
> pub/ASUSCOM/BIOS/Socket_A/AMD_Chipset/AMD_760_MPX/A7M266-D/
>
>

Asus A7M266-D, Linux 2.6.7 and APIC

[Andy Biddle]

I'm having what I suspect is a newbie-type problem, but I assure you I've
looked around.

I have a dual-proc server that I've recently decided to overhaul.  It's an
Asus A7M266-D motherboard.  It had been running with dual Athlon MP 1800+s
and RedHat and BIOS rev 1003 for at least a year with no real problems.

First I decided to change the OS to Gentoo.  I build the system with no
problem and built a custom kernel based on linux 2.6.7-gentoo-r11.
Everything was working great.  Then I changed out the procs and went with
dual Athlon MP 2800+s.  To support these, I needed to (according to Asus's
website) upgrade the BIOS to 1011.002 or higher.  (Latest is 1011.003, so
that's what I used.)

Now, when I boot to this custom kernel I get about 2 seconds into the boot
process before the system starts spewing constant "APIC error on CPU0:
04(04)" messages.   It stops booting at that point.

I've done a little research and it seems that I'm supposed to do a couple
of things:
	1.  Disable "MPS 1.4 Support" in the BIOS.
	2.  pass the kernel "noapic" as a parameter.

I've done both of these and I STILL get the APIC messages.  (Which leads
me to suspect the "noapic" isn't working or I'm doing it wrong at the very
least.)

I've build another custom kernel where the only difference is that SMP is
disabled.  Sure enough, that works like a champ, but with only the single
CPU.

I've searched through the kernel "make menuconfig" menus and can't see
that I'm missing anything.  In fact, I can't even FIND "APIC" options
unless I disable SMP...

I'm just about out of ideas here... Any suggestions?

Thanks,
AndyB
-
To unsubscribe from this list: send the line "unsubscribe linux-smp" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html