Re: 2.6.32.5 regression: page allocation failure. order:1,

[Mark Lord <kernel@teksavvy.com>]

Mark Lord wrote:
> Mark Lord wrote:
>> Mel Gorman wrote:
>>> On Tue, Jan 26, 2010 at 09:13:27PM -0500, Mark Lord wrote:
>>>> I recently upgraded our 24/7 server from 2.6.31.5 to 2.6.32.5.
>>>>
>>>> Now, suddenly the logs are full of "page allocation failure. order:1",
>>>> and the odd "page allocation failure. order:4" failures.
>>>>
>>>> Wow.  WTF happened in 2.6.32 ???
>>>>
>>>
>>> There was one bug related to MIGRATE_RESERVE that might be affecting
>>> you. It reported as impacting swap-orientated workloads but it could
>>> easily affect drivers that depend on high-order atomic allocations.
>>> Unfortunately, the fix is not signed-off yet but I expect it to make its
>>> way towards mainline when it is.
>>>
>>> Here is the patch with a slightly-altered changelog. Can you test if it
>>> makes a difference please?
>> ..
>>
>> We don't like to reboot our 24/7 server very often,
>> and certainly not for debugging buggy kernels.
>>
>> It's rock solid again with 2.6.31.12 on it now.
>>
>> The defining characteristic of that machine, is that it has only 512MB
>> of physical RAM.  So perhaps I'll try booting a different machine here
>> with mem=512M and see how that behaves.  If the problem shows up on that,
>> then I'll try the patch.
> ..
> 
> Sod it.  2.6.32 is simply too broken for us here on 32-bit non-SMP.
> 
> Attempting to boot a 32-bit kernel with "nosmp mem=512M" on my notebook
> locks up at boot time with several repeated messages like this:
> 
>   request_module: runaway loop modprobe binfmt_464c
> 
> Useless kernel on 32-bit.  I hope 2.6.33 ends up less buggy.
..

I rebuilt it (again!), this time as a pure UP (non-SMP) kernel,
and it still locks at boot, with or without the mem=512M parameter.

This is one really bad kernel release for 32-bit x86.

-ml