[BUG] Kernel Panic in squashfs

[BUG] Kernel Panic in squashfs

[Tim Tassonis]

Hi all


Just found out that squashfs panics when compiled in statically instead 
of as a module, when mounting an sqf file. The sequence I did was:


# mkdir /mnt/gaia-ro

# mount /gaiarule.sqf  /mnt/gaia-ro -t squashfs -o loop

Maybe it is of importance that the sqf file is located in the initramfs. 
The kernel was panicking reliably with different sqf files, on different 
hardware upon the mount command, with "unable to handle kernel paging 
request". This was on 3.18.5. As soon as I compiled squashfs as a 
modules, the problem went away.

If you need further details, please cc me directly, as I unsubscribed 
from the list due to not being able to handle the massive load.

Bye
Tim

Re: [BUG] Kernel Panic in squashfs

[Richard Weinberger]

On Fri, Feb 6, 2015 at 5:55 PM, Tim Tassonis <stuff@decentral.ch> wrote:
> Hi all
>
>
> Just found out that squashfs panics when compiled in statically instead of
> as a module, when mounting an sqf file. The sequence I did was:
>
>
> # mkdir /mnt/gaia-ro
>
> # mount /gaiarule.sqf  /mnt/gaia-ro -t squashfs -o loop
>
> Maybe it is of importance that the sqf file is located in the initramfs. The
> kernel was panicking reliably with different sqf files, on different
> hardware upon the mount command, with "unable to handle kernel paging
> request". This was on 3.18.5. As soon as I compiled squashfs as a modules,
> the problem went away.
>
> If you need further details, please cc me directly, as I unsubscribed from
> the list due to not being able to handle the massive load.

What about sharing the actual kernel panic? :-)

-- 
Thanks,
//richard

Re: [BUG] Kernel Panic in squashfs

[Tim Tassonis]

> On Fri, Feb 6, 2015 at 5:55 PM, Tim Tassonis <stuff@decentral.ch> wrote:
>> Hi all
>>
>>
>> Just found out that squashfs panics when compiled in statically instead of
>> as a module, when mounting an sqf file. The sequence I did was:
>>
>>
>> # mkdir /mnt/gaia-ro
>>
>> # mount /gaiarule.sqf  /mnt/gaia-ro -t squashfs -o loop
>>
>> Maybe it is of importance that the sqf file is located in the initramfs. The
>> kernel was panicking reliably with different sqf files, on different
>> hardware upon the mount command, with "unable to handle kernel paging
>> request". This was on 3.18.5. As soon as I compiled squashfs as a modules,
>> the problem went away.
>>
>> If you need further details, please cc me directly, as I unsubscribed from
>> the list due to not being able to handle the massive load.
>
> What about sharing the actual kernel panic? :-)

Well, I was trying to reproduce it. Upgraded to 3.18.6, built the kernel 
with static squashfs, and now it doesn't panic anymore.

Guess some fixes for 3.18.6 fixed this too. If somebody really, really 
cares, I could go through the whole procedure again with 3.18.6, but I 
guess, as it's gone now, this would be rather academeic.

Bye
Tim

Re: [BUG] Kernel Panic in squashfs

[Tim Tassonis]

> On Fri, Feb 6, 2015 at 5:55 PM, Tim Tassonis <stuff@decentral.ch> wrote:
>> Hi all
>>
>>
>> Just found out that squashfs panics when compiled in statically instead of
>> as a module, when mounting an sqf file. The sequence I did was:
>>
>>
>> # mkdir /mnt/gaia-ro
>>
>> # mount /gaiarule.sqf  /mnt/gaia-ro -t squashfs -o loop
>>
>> Maybe it is of importance that the sqf file is located in the initramfs. The
>> kernel was panicking reliably with different sqf files, on different
>> hardware upon the mount command, with "unable to handle kernel paging
>> request". This was on 3.18.5. As soon as I compiled squashfs as a modules,
>> the problem went away.
>>
>> If you need further details, please cc me directly, as I unsubscribed from
>> the list due to not being able to handle the massive load.
>
> What about sharing the actual kernel panic? :-)

Well, I will, if somebody bothers to actually look at it. Takes me some 
time to set everything up again for it. And it would be nice if you 
could cc me directly in the reply, as already mentioned, I'm not on the 
list and this copy-paste stuff from marc.info is a bit a nuisance.

Bye
Tim