Re: 3.6.11: khubd issue SLAB related?

[Udo van den Heuvel]

On 2013-03-11 22:23, David Rientjes wrote:
> On Fri, 8 Mar 2013, Udo van den Heuvel wrote:
> 
>> [3663200.799579] SLAB: Unable to allocate memory on node 0 (gfp=0xd0)
>> [3663200.799581]   cache: size-65536, object size: 65536, order: 4
>> [3663200.799583]   node 0: slabs: 6/6, objs: 6/6, free: 0
>> [3663200.799585] pwc: Oops, could not allocate memory for pwc_device.
> 
> You have memory freed, but like your earlier page allocation failure on 
> mount, it is too fragmented.  This is an order-4 page allocation, the 
> kmalloc() of struct pwc_device, and you don't even have any order-3 pages 
> that are available.  Are you running with CONFIG_COMPACTION enabled?

Yes.

> If not, enable it and try again, or get root and manually try to 
> defragment memory by doing echo 1 > /proc/sys/vm/compact_memory.

This needs to be done regularly, I assume?

> If so, try killing a memory hogging process to free some memory and 
> attempt to get some higher order pages available.

It's a 8GB box (7.5GB available to software) with modest load. Mostly
email & web browsing, some torrents, nntp stuff, etc. Nothing weird.
Around 4.8G in buffers/cache right now.

Why is this issue occurring?
Why did I not see this in the past?

Udo
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/