Bug#732279: linux-image-3.2.0-4-686-pae: netback page allocation failure (Xen memory leak?)

[Ben Hutchings <ben@decadent.org.uk>]

On Mon, Dec 16, 2013 at 09:41:56AM +0000, George B. wrote:
> Package: src:linux
> Version: 3.2.51-1
> Severity: normal
> 
> Hello,
> 
> I am seeing the backtrace below in my kernel log after the system has been running for several weeks.
> 
> Looks like it has something to do with Xen - memory leak maybe?

I'm not sure it's directly connected with Xen - I think that an
interrupt from the physical network interface interrupted a task that
is part of the netback driver.

This is not necessarily due to a memory leak; more likely this is
memory fragmentation.  Some oddity of your networking configuration
results in linearising large packets:

[...]
> [955212.368551] netback/0: page allocation failure: order:3, mode:0x20
[...]

This means: allocating something between 16 and 32K atomically
(no waiting allowed).

The call trace appears to show that a packet received by the local TCP
via a bridge resulted in an immediate transmission, again going
through the bridge, and that then required this large memory
allocation.

Please provide details of your networking configuration,
including:

- Are you using ebtables?
- Are you using VLAN devices?

Ben.

-- 
Ben Hutchings
Life is like a sewer:
what you get out of it depends on what you put into it.