2.6.13 does not boot on Eiger

2.6.13 does not boot on Eiger

[Andreas Schwab]

[-- Attachment #1: Type: text/plain, Size: 1008 bytes --]

I'm getting an MCA during early setup on Eiger, these are the last (and
only) messages:

EFI v1.10 by HP: SALsystab=0x723feee9710 ACPI 2.0=0x723fe040000
HCDP=0x723fe0720d0 SMBIOS=0x7fffe000
booting generic kernel on platform hpzx1
PCDP: v1 at 0x723fe0720d0
Early serial console at MMIO 0xf0100005000 (options '9600n8')
Number of logical nodes in system = 3
Number of memory chunks in system = 34
warning: skipping physical page 0
warning: skipping physical page 0
warning: skipping physical page 0

CONFIG_IA64_UNCACHED_ALLOCATOR is already disabled, and the kernel works
fine on Tiger, so there must be some different problem.  The config is
attached.  Note that this is already broken since at least 2.6.13-rc3, but
2.6.13-rc1 ist still ok.

Andreas.

-- 
Andreas Schwab, SuSE Labs, schwab@suse.de
SuSE Linux Products GmbH, Maxfeldstraße 5, 90409 Nürnberg, Germany
Key fingerprint = 58CA 54C7 6D53 942B 1756  01D3 44D5 214B 8276 4ED5
"And now for something completely different."

[-- Attachment #2: config.gz --]
[-- Type: application/x-gzip, Size: 13838 bytes --]

Re: 2.6.13 does not boot on Eiger

[Bjorn Helgaas]

On Tuesday 30 August 2005 5:04 pm, Andreas Schwab wrote:
> I'm getting an MCA during early setup on Eiger, these are the last (and
> only) messages:
> 
> EFI v1.10 by HP: SALsystab=0x723feee9710 ACPI 2.0=0x723fe040000
> HCDP=0x723fe0720d0 SMBIOS=0x7fffe000
> booting generic kernel on platform hpzx1
> PCDP: v1 at 0x723fe0720d0
> Early serial console at MMIO 0xf0100005000 (options '9600n8')
> Number of logical nodes in system = 3
> Number of memory chunks in system = 34
> warning: skipping physical page 0
> warning: skipping physical page 0
> warning: skipping physical page 0
> 
> CONFIG_IA64_UNCACHED_ALLOCATOR is already disabled, and the kernel works
> fine on Tiger, so there must be some different problem.  The config is
> attached.  Note that this is already broken since at least 2.6.13-rc3, but
> 2.6.13-rc1 ist still ok.

Does the MCA log say anything useful?

I used your .config and saw an MCA in i8042_flush() (IIP in
__ia64_inb, BR0 in i8042_flush).  Can you try turning off
CONFIG_SERIO_I8042?  This was fixed a while back, but maybe
it got broken again.  And the i8042 probe happens much later
than your log indicates, so maybe you're seeing a different
problem.

P.S.  What is your firmware "acpiconfig" mode?  My eiger is
in "default" mode, in which accesses to non-existent I/O ports
cause MCAs.  But we don't officially support that yet because
of issues like this.  The supported modes allow accesses to
non-existent ports by setting either "single-pci-domain" or
"enable legacyprobe".

Re: 2.6.13 does not boot on Eiger

[Andreas Schwab]

Bjorn Helgaas <bjorn.helgaas@hp.com> writes:

> Does the MCA log say anything useful?

How can I find out?

> P.S.  What is your firmware "acpiconfig" mode? My eiger is
> in "default" mode,

Likewise.

> The supported modes allow accesses to non-existent ports by setting
> either "single-pci-domain" or "enable legacyprobe".

Neither seem to be supported here.

Andreas.

-- 
Andreas Schwab, SuSE Labs, schwab@suse.de
SuSE Linux Products GmbH, Maxfeldstraße 5, 90409 Nürnberg, Germany
Key fingerprint = 58CA 54C7 6D53 942B 1756  01D3 44D5 214B 8276 4ED5
"And now for something completely different."

Re: 2.6.13 does not boot on Eiger

[Bjorn Helgaas]

On Wednesday 31 August 2005 10:03 am, Bjorn Helgaas wrote:
> I used your .config and saw an MCA in i8042_flush() (IIP in
> __ia64_inb, BR0 in i8042_flush).  Can you try turning off
> CONFIG_SERIO_I8042?  This was fixed a while back, but maybe
> it got broken again.

Oh, I remember now.  I debugged this a while back.  The i8042
fix was originally done in terms of ACPI, but recently changed
to be PNP-based.  So on ia64, we need CONFIG_PNP and CONFIG_PNPACPI.

These are turned off in your .config (and in all the configs in
the tree).  I'll post a patch to fix those.

Re: 2.6.13 does not boot on Eiger

[Bjorn Helgaas]

On Wednesday 31 August 2005 10:18 am, Andreas Schwab wrote:
> Bjorn Helgaas <bjorn.helgaas@hp.com> writes:
> 
> > Does the MCA log say anything useful?
> 
> How can I find out?

On HP boxes, you can do "errdump mca" (for zx1 boxes) or
"errdump all mca" (for sx1000 boxes) at the EFI shell.
If you can boot a working kernel, salinfod will also
extract the logs and drop them in /var/log/salinfo.
There's a decoded/ directory, which is basically the
same output as "errdump", and a raw/ directory, which
contains binary files that can be run through a decoder
(which is unfortunately proprietary).

> > P.S.  What is your firmware "acpiconfig" mode? My eiger is
> > in "default" mode,
> 
> Likewise.

Type "acpiconfig" at the EFI shell prompt.

Re: 2.6.13 does not boot on Eiger

[Andreas Schwab]

Bjorn Helgaas <bjorn.helgaas@hp.com> writes:

> If you can boot a working kernel, salinfod will also
> extract the logs and drop them in /var/log/salinfo.
> There's a decoded/ directory, which is basically the
> same output as "errdump",

They all say: "UNCORRECTED PROCESSOR ERROR: Cache Check".  I don't know
what is needed from the files to make this useful.

>> > P.S.  What is your firmware "acpiconfig" mode? My eiger is
>> > in "default" mode,
>> 
>> Likewise.
>
> Type "acpiconfig" at the EFI shell prompt.

"Likewise" was actually referring to "default".

Andreas.

-- 
Andreas Schwab, SuSE Labs, schwab@suse.de
SuSE Linux Products GmbH, Maxfeldstraße 5, 90409 Nürnberg, Germany
Key fingerprint = 58CA 54C7 6D53 942B 1756  01D3 44D5 214B 8276 4ED5
"And now for something completely different."