Am 19.10.2010 schrieb "Jan Beulich <JBeulich@xxxxxxxxxx>":
> >>> On 19.10.10 at 14:16, Dietmar Hahn <dietmar.hahn@xxxxxxxxxxxxxx> wrote:
> > (XEN) Xen BUG at page_alloc.c:385
> > The code line is:
> > BUG_ON(pg[i].count_info != PGC_state_free);
> > Is it possible in any way that a free page is in use? Or is it always a
> > failure somewhere in the code like a multiple use of the same page?
> Free pages are never in use for anything, so getting this BUG_ON()
> to trigger means e.g. something modified a page's state after it was
> already passed back to the allocator.
> Did you check that this is not another manifestation of the problem
> fixed by c/s 22135 (we have a backport of this in our tree, but it's
> not clear whether yours already has this).
Sorry I wasn't aware of this fix and the backport is not in our source.
Company details: http://ts.fujitsu.com/imprint.html