Re: [PATCH 1/3] slqb: Do not use DEFINE_PER_CPU for per-node data

[Tejun Heo <tj@kernel.org>]

Hello,

Mel Gorman wrote:
> On Mon, Sep 21, 2009 at 06:00:22PM +0900, Tejun Heo wrote:
>> Hello,
>>
>> Mel Gorman wrote:
>>>>> Can you please post full dmesg showing the corruption? 
>>> There isn't a useful dmesg available and my evidence that it's within the
>>> pcpu allocator is a bit weak.
>> I'd really like to see the memory layout, especially how far apart the
>> nodes are.
>>
> 
> Here is the console log with just your patch applied. The node layouts
> are included in the log although I note they are not far apart. What is
> also important is that the exact location of the bug is not reliable
> although it's always in accessing the same structure. This time it was a
> bad data access. The time after that, a BUG_ON triggered when locking a
> spinlock in the same structure. The third time, it locked up silently.
> Forth time, it was a data access error but a different address and so
> on.

One likely possibility is something accessing wrong percpu offset.
Can you please attach .config?

Thanks.

-- 
tejun