Sun v440 and 2.6.13-1.1552sp1smp (Aurora) kernel

Sun v440 and 2.6.13-1.1552sp1smp (Aurora) kernel

[Tomasz Kłoczko]

[-- Attachment #1: Type: TEXT/PLAIN, Size: 854 bytes --]


Still no progress ..

Boot device: /pci@1f,700000/scsi@2/disk@0,0:a  File and args:
SILO Version 1.4.8
boot:
linux
boot: linux -p
Allocated 8 Megs of memory at 0x40000000 for kernel
Loaded kernel version 2.6.13
Loading initial ramdisk (705508 bytes at 0x333F800000 phys, 0x40C00000 
virt)...
/
PROMLIB: Sun IEEE Boot Prom 4.13.0 2004/01/19 18:28
Linux version 2.6.13-1.1552sp1smp (root@arthur.devel.redhat.com) (gcc version 3.4.2 20041017 (Red Hat 3.4.2-6.fc3)) #1 SMP Tue Sep 13 14:22:30 EDT 2005
ARCH: SUN4U
Remapping the kernel... done.

.. and that is all.

kloczek
-- 
-----------------------------------------------------------
*Ludzie nie mają problemów, tylko sobie sami je stwarzają*
-----------------------------------------------------------
Tomasz Kłoczko, sys adm @zie.pg.gda.pl|*e-mail: kloczek@rudy.mif.pg.gda.pl*

Re: Sun v440 and 2.6.13-1.1552sp1smp (Aurora) kernel

[Tomasz 'Furgas' Sawicki]

Tomasz K³oczko wrote:
> Remapping the kernel... done.
> 
> .. and that is all.

I have the same effect on v240. Booting from network works like a charm,
but booting from SILO stops after "Remapping the kernel... done". I'm
using kernel 2.4.26smp.
I reported this problem here more than a year ago:
http://marc.theaimsgroup.com/?l=linux-sparc&m\x108639201906499&w=2

-- 
Tomasz 'Furgas' Sawicki
Computer Network Management Centre
Poznan University of Technology

Re: Sun v440 and 2.6.13-1.1552sp1smp (Aurora) kernel

[David S. Miller]

From: "Tomasz 'Furgas' Sawicki" <tomek@sawka.prv.pl>
Date: Wed, 14 Sep 2005 08:19:41 +0200

> I reported this problem here more than a year ago:
> http://marc.theaimsgroup.com/?l=linux-sparc&m\x108639201906499&w=2

The problem is that the people who can work on fixing
this problem (whether it is in the kernel or SILO,
we still don't know) are not able to reproduce it.

Re: Sun v440 and 2.6.13-1.1552sp1smp (Aurora) kernel

[Tomasz Kłoczko]

[-- Attachment #1: Type: TEXT/PLAIN, Size: 752 bytes --]

On Wed, 14 Sep 2005, Tomasz 'Furgas' Sawicki wrote:

> Tomasz Kłoczko wrote:
>> Remapping the kernel... done.
>> 
>> .. and that is all.
>
> I have the same effect on v240. Booting from network works like a charm,

Using TILO or other method ?

> but booting from SILO stops after "Remapping the kernel... done". I'm
> using kernel 2.4.26smp.
> I reported this problem here more than a year ago:
> http://marc.theaimsgroup.com/?l=linux-sparc&m=108639201906499&w=2

kloczek
-- 
-----------------------------------------------------------
*Ludzie nie mają problemów, tylko sobie sami je stwarzają*
-----------------------------------------------------------
Tomasz Kłoczko, sys adm @zie.pg.gda.pl|*e-mail: kloczek@rudy.mif.pg.gda.pl*

Re: Sun v440 and 2.6.13-1.1552sp1smp (Aurora) kernel

[Tomasz 'Furgas' Sawicki]

Tomasz K³oczko wrote:
> On Wed, 14 Sep 2005, Tomasz 'Furgas' Sawicki wrote:
> 
>> Tomasz K³oczko wrote:
>>
>>> Remapping the kernel... done.
>>>
>>> .. and that is all.
>>
>>
>> I have the same effect on v240. Booting from network works like a charm,
> 
> 
> Using TILO or other method ?

Using PROM built-in option: boot net:dhcp.

-- 
Tomasz 'Furgas' Sawicki
Computer Network Management Centre
Poznan University of Technology

Re: Sun v440 and 2.6.13-1.1552sp1smp (Aurora) kernel

[Gustavo Zacarias]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

David S. Miller wrote:

> The problem is that the people who can work on fixing
> this problem (whether it is in the kernel or SILO,
> we still don't know) are not able to reproduce it.

The problem lies in SILO, we're using 1.2.6 in our (gentoo) v240 just
fine to boot off disk.
No kernel is rock-solid stable on the v240, though 2.4 lasts longer.
I'm guessing it has to do with IIIi SMP since we get tons of "CPU[1]:
Cheetah+ D-cache parity error at TPC[00000000005d3328]" (with varying
CPU number and addresses) before one of the processors goes offline/gets
stuck at which point it's time to reboot.

- --
Gustavo Zacarias
Gentoo/SPARC monkey
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.1 (GNU/Linux)
Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org

iD8DBQFDKETlV3G/IBCn/JARAhqdAJ4/Y+Y7Q+hLqWVteH96dSM+6j1BTwCfZ+8Y
9BfFesOsGyNf5wzGYkpqwvM=FlfU
-----END PGP SIGNATURE-----

Re: Sun v440 and 2.6.13-1.1552sp1smp (Aurora) kernel

[David S. Miller]

From: Gustavo Zacarias <gustavoz@gentoo.org>
Date: Wed, 14 Sep 2005 12:42:29 -0300

> No kernel is rock-solid stable on the v240, though 2.4 lasts longer.
> I'm guessing it has to do with IIIi SMP since we get tons of "CPU[1]:
> Cheetah+ D-cache parity error at TPC[00000000005d3328]" (with varying
> CPU number and addresses) before one of the processors goes offline/gets
> stuck at which point it's time to reboot.

Known SMP bug with UltraSPARC-IIIi processors.  They can randomly hang
during an L2 cache transaction, and the only way to get them unstuck
is to gain ownership of an L2 cache line (from the other cpu which is
still alive) that stuck processor holds.

I'm not willing to work on a workaround for this chip errata until
I have physical access to an SMP UltraSPARC-IIIi system for several
weeks.

Re: Sun v440 and 2.6.13-1.1552sp1smp (Aurora) kernel

[Peter Jones]

On Wed, 2005-09-14 at 12:42 -0300, Gustavo Zacarias wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
> 
> David S. Miller wrote:
> 
> > The problem is that the people who can work on fixing
> > this problem (whether it is in the kernel or SILO,
> > we still don't know) are not able to reproduce it.
> 
> The problem lies in SILO, we're using 1.2.6 in our (gentoo) v240 just
> fine to boot off disk.

The fact that an older version of SILO works does not actually mean the
problem is in SILO.  SILO writes to memory, kernel reads from that
memory.  Either part can be buggy, and both have in the past.

If people want this fixed, they need to get a box that's breaking in
front of one of those of us who can debug this sort of failure or debug
it themselves.

-- 
  Peter