Re: [ 07/48] mm: fix slab->page _count corruption when using slub

[Ben Hutchings]

On Mon, 2012-07-02 at 20:46 -0300, Herton Ronaldo Krzesinski wrote:
> On Sun, Jul 01, 2012 at 06:20:13PM +0100, Ben Hutchings wrote:
> > 3.2-stable review patch.  If anyone has any objections, please let me know.
> > 
> > ------------------
> > 
> > From: Pravin B Shelar <pshelar@xxxxxxxxxx>
> > 
> > commit abca7c4965845924f65d40e0aa1092bdd895e314 upstream.
> > 
> > On arches that do not support this_cpu_cmpxchg_double() slab_lock is used
> > to do atomic cmpxchg() on double word which contains page->_count.  The
> > page count can be changed from get_page() or put_page() without taking
> > slab_lock.  That corrupts page counter.
> > 
> > Fix it by moving page->_count out of cmpxchg_double data.  So that slub
> > does no change it while updating slub meta-data in struct page.
> 
> Hi,
> 
> I got a frozen machine on boot with 3.2.22 proposed patches, bisect
> pointed out to this commit (and reverting only this commit on top of
> 3.2.22 patches confirmed it). So far I was able to reproduce always.
[...]

Sorry, I should remember to test with SLUB whenever there's a patch that
touches it!  I'll drop this for now, given Pravin's feedback.

Please let me know when this is completely fixed in mainline and ready
to be cherry-picked/backported.

Ben.

-- 
Ben Hutchings
When in doubt, use brute force. - Ken Thompson

Attachment: signature.asc
Description: This is a digitally signed message part