[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Bug#666017: marked as done (linux-image-3.2.0-0.bpo.2-686-pae crashes during or after initial ramdisk on alix6e2 hardware)

Your message dated Thu, 29 Mar 2012 15:26:00 +0100
with message-id <1333031160.3500.294.camel@deadeye>
and subject line Re: Bug#666017: linux-image-3.2.0-0.bpo.2-686-pae crashes during or after initial ramdisk on alix6e2 hardware
has caused the Debian Bug report #666017,
regarding linux-image-3.2.0-0.bpo.2-686-pae crashes during or after initial ramdisk on alix6e2 hardware
to be marked as done.

This means that you claim that the problem has been dealt with.
If this is not the case it is now your responsibility to reopen the
Bug report if necessary, and/or fix the problem forthwith.

(NB: If you are a system administrator and have no idea what this
message is talking about, this may indicate a serious mail system
misconfiguration somewhere. Please contact owner@bugs.debian.org
immediately.)


-- 
666017: http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=666017
Debian Bug Tracking System
Contact owner@bugs.debian.org with problems
--- Begin Message --- 
Package: linux-image-3.2.0-0.bpo.2-686-pae
Version: 3.2.12-1~bpo60+1
Severity: important
Tags: upstream

Dear Maintainer,

The system crashes during boot:

# the grub-pc options
linux /boot/vmlinuz-3.2.0-0.bpo.2-686-pae root=UUID=ce0d589d-121d-449b-9fc9-7e392b17fbda ro single console=ttyS0,115200

# the console output:
Loading Linux 3.2.0-0.bpo.2-686-pae ...
Loading initial ramdisk ...
PC Engines ALIX.2 v0.99h
640 KB Base Memory
211968

The text PC Engines ALIX.2 incates it rebooted itself.

I have been using Debian 2.6 kernels successful for many years
hardware: http://www.pcengines.ch/alix6e2.htm

Installed a 2.6 kernel and the system booted:

# works!
BOOT_IMAGE=/boot/vmlinuz-2.6.32-5-686 root=UUID=ce0d589d-121d-449b-9fc9-7e392b17fbda ro console=ttyS0,115200 quiet

I ran a Memtest86+ v4.10 to be sure the HW is OK:
*****Pass complete, no errors, press Esc to exit*****

I want the 3.x kernels to work on these systems, so I can keep ketting new features and security updates...

PS. I made this report from my workstation

-- System Information:
Debian Release: wheezy/sid
  APT prefers testing
  APT policy: (990, 'testing'), (500, 'unstable'), (500, 'stable'), (1, 'experimental')
Architecture: amd64 (x86_64)

Kernel: Linux 3.2.0-2-amd64 (SMP w/4 CPU cores)
Locale: LANG=en_GB.UTF-8, LC_CTYPE=en_GB.UTF-8 (charmap=UTF-8)
Shell: /bin/sh linked to /bin/dash

--- End Message ---
--- Begin Message --- 
On Thu, 2012-03-29 at 13:12 +0100, Simon McVittie wrote:
> On Thu, 29 Mar 2012 at 13:30:54 +0200, Jelle de Jong wrote:
> > When will there be a non pea 3.x kernel like with the 2.6 kernel
> > available in Debian?
> 
> >From the kernel changelog:
> 
>    * [i386] Rename '686-bigmem' flavour to '686-pae'; remove '686' flavour.
>      For 686-class systems without PAE, the '486' flavour is more efficient
>      than the '686' flavour due to optimisation for uniprocessor systems.
> 
> So you should use the -486 flavour on non-PAE hardware. Larger numbers
> aren't necessarily better :-)
> 
> Perhaps the linux-image-686 transitional package should depend on
> linux-image-486 (the conservative/safe/lowest-common-denominator choice).

The transitional packages linux-image-686 and linux-image-2.6-686
include a config script that will abort and explain the situation if the
system does not have PAE.  Jelle apparently didn't install one of them.

(It's a shame that Linux doesn't check CPU flags and print an explicit
error message at boot time, but I think there's a bootstrap problem
there.  It won't have a proper console driver available until after it's
set up page tables, so you wouldn't see the message unless you set
earlyprintk.)

> On the other hand, linux-image-686-pae (the former "bigmem" version) is
> more appropriate for most mainstream CPUs (although not for unusual CPUs like
> your Geode), which is presumably the reason why automatic migration
> from -686 is to that one.

I think that the large majority of systems using the '686' flavour have
PAE.  Many (most?) are SMP so the '486' flavour would be a large
regression.

> The kernel team's advice on the appropriate new kernel for various CPUs
> should probably go in the wheezy release notes, particularly if cases
> like this exist.

There should perhaps be a note about upgrading so that users who need to
switch from 686 to 486 can do that at the start of the upgrade (avoiding
an error in the middle).  For any other upgrade and new installations,
It Just Works.

Ben.

-- 
Ben Hutchings
Horngren's Observation:
                   Among economists, the real world is often a special case.

Attachment: signature.asc
Description: This is a digitally signed message part


--- End Message ---
Reply to: