WARNING: MP table in the EBDA can be UNSAFE

WARNING: MP table in the EBDA can be UNSAFE

[Paul Rolland]

Hello,

Our Linux, running on an IBM X-Series 445, says :
(excerpt from dmesg) :

Linux version 2.4.20 (root@stg-02.ntr.witbe.francetelecom.fr) (gcc version 2.96 20000731 (Red Hat Linux 7.1 2.96-98)) #12 SMP
Fri Dec 5 11:08:40 GMT 2003
BIOS-provided physical RAM map:
 BIOS-e820: 0000000000000000 - 000000000009dc00 (usable)
 BIOS-e820: 000000000009dc00 - 00000000000a0000 (reserved)
 BIOS-e820: 00000000000e0000 - 0000000000100000 (reserved)
 BIOS-e820: 0000000000100000 - 000000007ffb7000 (usable)
 BIOS-e820: 000000007ffb7000 - 000000007ffbf800 (ACPI data)
 BIOS-e820: 000000007ffbf800 - 0000000080000000 (reserved)
 BIOS-e820: 00000000fec00000 - 0000000100000000 (reserved)
1151MB HIGHMEM available.
896MB LOWMEM available.
found SMP MP-table at 0009dd40
hm, page 0009d000 reserved twice.
hm, page 0009e000 reserved twice.
hm, page 0009e000 reserved twice.
hm, page 0009f000 reserved twice.
WARNING: MP table in the EBDA can be UNSAFE, contact linux-smp@vger.kernel.org if you experience SMP problems!
On node 0 totalpages: 524215
zone(0): 4096 pages.
zone(1): 225280 pages.
zone(2): 294839 pages.
ACPI: Searched entire block, no RSDP was found.
ACPI: RSDP located at physical address c00fdba0
RSD PTR  v0 [IBM   ]
__va_range(0x7ffbf780, 0x68): idx=8 mapped at ffff6000
ACPI table found: RSDT v1 [IBM    SERVIGIL 0.4096]
__va_range(0x7ffbf700, 0x24): idx=8 mapped at ffff6000
__va_range(0x7ffbf700, 0x74): idx=8 mapped at ffff6000
ACPI table found: FACP v1 [IBM    SERVIGIL 0.4096]
__va_range(0x7ffbf640, 0x24): idx=8 mapped at ffff6000
__va_range(0x7ffbf640, 0x9a): idx=8 mapped at ffff6000
ACPI table found: APIC v1 [IBM    SERVIGIL 0.4096]
__va_range(0x7ffbf640, 0x9a): idx=8 mapped at ffff6000
LAPIC (acpi_id[0x0000] id[0x0] enabled[1])
CPU 0 (0x0000) enabledProcessor #0 Pentium 4(tm) XEON(tm) APIC version 16
 
LAPIC (acpi_id[0x0001] id[0x12] enabled[1])
CPU 1 (0x1200) enabled<4>Processor #18 INVALID - (Max ID: 16).
LAPIC (acpi_id[0x0002] id[0x20] enabled[1])
CPU 1 (0x2000) enabled<4>Processor #32 INVALID - (Max ID: 16).
LAPIC (acpi_id[0x0003] id[0x32] enabled[1])
CPU 1 (0x3200) enabled<4>Processor #50 INVALID - (Max ID: 16).

Is there any known solution to re-enable the CPU 1, CPU 2 and CPU 3 ?

Is this an IBM bug ?

Please reply privately, I'm not subscribed to linux-smp.

Thanks in advance,
Paul

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Blessed are they who have nothing to say and cannot be persuaded to say it.

-James Russell Lowell
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~


Paul Rolland, rol@witbe.net
Witbe.net SA
Directeur Associe

--

Please no HTML, I'm not a browser - Pas d'HTML, je ne suis pas un navigateur

"Some people dream of success... while others wake up and work hard at it"

Re: WARNING: MP table in the EBDA can be UNSAFE

[Zwane Mwaikambo]

On Fri, 5 Dec 2003, Paul Rolland wrote:

> Our Linux, running on an IBM X-Series 445, says :
> (excerpt from dmesg) :
>
> found SMP MP-table at 0009dd40
> hm, page 0009d000 reserved twice.
> hm, page 0009e000 reserved twice.
> hm, page 0009e000 reserved twice.
> hm, page 0009f000 reserved twice.
> WARNING: MP table in the EBDA can be UNSAFE, contact linux-smp@vger.kernel.org if you experience SMP problems!

This bit is ok and can be safely ignored.

> LAPIC (acpi_id[0x0000] id[0x0] enabled[1])
> CPU 0 (0x0000) enabledProcessor #0 Pentium 4(tm) XEON(tm) APIC version 16
>
> LAPIC (acpi_id[0x0001] id[0x12] enabled[1])
> CPU 1 (0x1200) enabled<4>Processor #18 INVALID - (Max ID: 16).
> LAPIC (acpi_id[0x0002] id[0x20] enabled[1])
> CPU 1 (0x2000) enabled<4>Processor #32 INVALID - (Max ID: 16).
> LAPIC (acpi_id[0x0003] id[0x32] enabled[1])
> CPU 1 (0x3200) enabled<4>Processor #50 INVALID - (Max ID: 16).
>
> Is there any known solution to re-enable the CPU 1, CPU 2 and CPU 3 ?
>
> Is this an IBM bug ?

Did you compile your kernel with the following option?
IBM x440 Summit/EXA support

CONFIG_X86_SUMMIT

Re: WARNING: MP table in the EBDA can be UNSAFE

[Paul Rolland]

Hello Zwane,

> > WARNING: MP table in the EBDA can be UNSAFE, contact 
> linux-smp@vger.kernel.org if you experience SMP problems!
> 
> This bit is ok and can be safely ignored.
OK, so let's forget about this one.
 
> Did you compile your kernel with the following option?
> IBM x440 Summit/EXA support
> 
> CONFIG_X86_SUMMIT
> 
I can't find this option ? Is this part of the 2.4.x branch ?

Paul

Re: WARNING: MP table in the EBDA can be UNSAFE

[Zwane Mwaikambo]

On Fri, 5 Dec 2003, Paul Rolland wrote:

> > Did you compile your kernel with the following option?
> > IBM x440 Summit/EXA support
> >
> > CONFIG_X86_SUMMIT
> >
> I can't find this option ? Is this part of the 2.4.x branch ?

Indeed it is, you need to turn on "Multi-node NUMA system support"

CONFIG_X86_NUMA

Re: WARNING: MP table in the EBDA can be UNSAFE

[Paul Rolland]

Hello,

> Indeed it is, you need to turn on "Multi-node NUMA system support"
> 
> CONFIG_X86_NUMA
> 
Gee... I was in a 2.4.20 kernel, and the SUMMIT option is not present,
but it is in 2.4.23...

I'll have to go to the server and do the re-installation using this
solution. I'll tell you Monday if it's working this way.

Regards, and many thanks for the tip,
Paul

Re: WARNING: MP table in the EBDA can be UNSAFE

[Paul Rolland]

Hello,

> > > Did you compile your kernel with the following option?
> > > IBM x440 Summit/EXA support
> > >
> > > CONFIG_X86_SUMMIT
> > >
> > I can't find this option ? Is this part of the 2.4.x branch ?
> 
> Indeed it is, you need to turn on "Multi-node NUMA system support"
> 
> CONFIG_X86_NUMA

That was it !

Great, the machine now sees all its CPUs.

The BogoMips number is rather strange (200 for a Xeon 2.4 GHz), but
the machine is really fine now....

Regards,
Paul

Re: WARNING: MP table in the EBDA can be UNSAFE

[Zwane Mwaikambo]

On Mon, 8 Dec 2003, Paul Rolland wrote:

> > Indeed it is, you need to turn on "Multi-node NUMA system support"
> >
> > CONFIG_X86_NUMA
>
> That was it !
>
> Great, the machine now sees all its CPUs.
>
> The BogoMips number is rather strange (200 for a Xeon 2.4 GHz), but
> the machine is really fine now....

I wonder if that's an artifact of the timesource, for my edification could
you send your new bootlogs?

Ta