SMP on Compaq DL380

[Matthew Sullivan]

Andre Guibert de Bruet wrote:

>
> On Sat, 23 Apr 2005, Matthew Sullivan wrote:
>
>> Andre Guibert de Bruet wrote:
>>
>>> On Sat, 23 Apr 2005, Matthew Sullivan wrote:
>>>
>>>> Andre Guibert de Bruet wrote:
>>>>
>>>>> On Thu, 21 Apr 2005, Matthew Sullivan wrote:
>>>>>
>>>>>> I've been reading about problems with HP/Compaq's regarding 
>>>>>> launching of second CPUs on SMP systems.
>>>>>>
>>>>>> I've been through the BIOS settings and there seems to be no 
>>>>>> settings to change the APCI table etc....
>>>>>>
>>>>>> Now one thing that does seem common, when I have BIOS's with MP 
>>>>>> table version set to 1.4 FreeBSD doesn't report the second CPU 
>>>>>> being launched (even though it is seen in the acpidump)....  When 
>>>>>> I set the BIOS to version 1.2 of the MP table the second CPU is 
>>>>>> reported and launched.
>>>>>>
>>>>>> Now the Compaq DL380's I have done seem to have the ability to 
>>>>>> set 1.4 or 1.2 of the table ... mptable reports 1.4... (below)
>>>>>>
>>>>>> Any suggestions on how to launch the second CPU...?
>>>>>
>>>>>
>>>>> Make a boot -v from this machine available.
>>>>
>>>>
>>>> http://scorpion.sorbs.net/dmesg.txt
>>>
>>>
>>> The lack of the following seems to indicate that you do not have 
>>> "device apic" enabled in your kernel config (You need "options SMP" 
>>> as well to get FreeBSD to do more than just recognize both CPUs):
>>>
>>> FreeBSD/SMP: Multiprocessor System Detected: 2 CPUs
>>>  cpu0 (BSP): APIC ID:  0
>>>  cpu1 (AP): APIC ID:  <some number>
>>>
>>> Please share your config and the steps that you are taking to build 
>>> your kernel.
>>
>>
>> /usr/src/sys/i386/conf/SCORPION has been copied to:
>> http://scorpion.sorbs.net/SCORPION
>>
>> /etc/make.conf contains 'KERNCONF=SCORPION'
>> then I follow the instructions in the Makefile....
>>
>> cd /usr/src
>> make buildworld
>> make buildkernel
>> make installkernel
>> reboot
>> mergemaster -p
>> make installworld
>> mergemaster
>> reboot
>>
>> (before I read the man page for make.conf I was using 
>> KERNCONF=SCORPION in the appropriate places on the command line)
>
>
> The dmesg shows that you compiled the kernel using this config file 
> anyway. All is good so far.
>
> Processors:     APIC ID Version State           Family  Model   
> Step    Flags
>         0       0x10    BSP, usable     6       2       1       0x0381
>         0       0x10    AP, usable      6       8       6       0x383fbff
>
> The APIC IDs here are the same. The flags on the would-be AP are what 
> I would expect for a recent i686. The BSP barely qualify it to be a 
> gen-1 Pentium. I wouldn't trust any of the values being reported. 
> Could you obtain the real identity of these CPUs and confirm that 
> they're not mismatched? The easy way of doing this if your BIOS 
> doesn't post this information is using a Knoppix LiveCD and doing a 
> cat /proc/cpuinfo.

Ok can't do the knoppix thing atm, however...

CPU0 -> 866/256/133/1.65v SL47S
CPU1 -> 866/256/133/1.70v SL48V

Both are shown detected by the BIOS, and both are shown as 866MHz 133MHz 
busses, and 256k cache (as one would expect)

>
> If both CPUs are reporting the same ID, I can see how we're not 
> launching the second proc; We assume that ID 0 is the BSP and 
> additional processors have different APIC IDs. Is something really 
> borked here? Yep!

But the acpidump -t shows 2 different ID's....

Regards,

Mat


-- 
Matthew Sullivan
Specialist Systems Programmer
Information Technology Services
The University of Queensland

-------------- next part --------------
A non-text attachment was scrubbed...
Name: smime.p7s
Type: application/x-pkcs7-signature
Size: 2666 bytes
Desc: S/MIME Cryptographic Signature
Url : http://lists.freebsd.org/pipermail/freebsd-current/attachments/20050424/b5b7d9e4/smime.bin