Re: [PATCH 06/19] mm: store mapcount for compound page separate

From: "Kirill A. Shutemov" <kirill@shutemov.name>
To: "Aneesh Kumar K.V" <aneesh.kumar@linux.vnet.ibm.com>
Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com>,
	"Kirill A. Shutemov" <kirill.shutemov@linux.intel.com>,
	Andrew Morton <akpm@linux-foundation.org>,
	Andrea Arcangeli <aarcange@redhat.com>,
	Dave Hansen <dave.hansen@intel.com>,
	Hugh Dickins <hughd@google.com>, Mel Gorman <mgorman@suse.de>,
	Rik van Riel <riel@redhat.com>, Vlastimil Babka <vbabka@suse.cz>,
	Christoph Lameter <cl@gentwo.org>,
	Steve Capper <steve.capper@linaro.org>,
	Johannes Weiner <hannes@cmpxchg.org>,
	Michal Hocko <mhocko@suse.cz>,
	"linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>,
	"linux-mm@kvack.org" <linux-mm@kvack.org>
Subject: Re: [PATCH 06/19] mm: store mapcount for compound page separate
Date: Fri, 21 Nov 2014 13:47:09 +0200	[thread overview]
Message-ID: <20141121114709.GA16647@node.dhcp.inet.fi> (raw)
In-Reply-To: <87egsx6oo1.fsf@linux.vnet.ibm.com>

On Fri, Nov 21, 2014 at 12:11:34PM +0530, Aneesh Kumar K.V wrote:
> "Kirill A. Shutemov" <kirill@shutemov.name> writes:
> 
> > On Tue, Nov 18, 2014 at 08:43:00AM +0000, Naoya Horiguchi wrote:
> >> > @@ -1837,6 +1839,9 @@ static void __split_huge_page_refcount(struct page *page,
> >> >  	atomic_sub(tail_count, &page->_count);
> >> >  	BUG_ON(atomic_read(&page->_count) <= 0);
> >> >  
> >> > +	page->_mapcount = *compound_mapcount_ptr(page);
> >> 
> >> Is atomic_set() necessary?
> >
> > Do you mean
> > 	atomic_set(&page->_mapcount, atomic_read(compound_mapcount_ptr(page)));
> > ?
> >
> > I don't see why we would need this. Simple assignment should work just
> > fine. Or we have archs which will break?
> 
> Are you looking at architecture related atomic_set issues, or the fact
> that we cannot have parallel _mapcount update and hence the above
> assignment should be ok ? If the former, current thp code
> use atomic_add instead of even using atomic_set when
> updatinge page_tail->_count.  
> 
> 		 * from under us on the tail_page. If we used
> 		 * atomic_set() below instead of atomic_add(), we
> 		 * would then run atomic_set() concurrently with
> 		 * get_page_unless_zero(), and atomic_set() is
> 		 * implemented in C not using locked ops. spin_unlock
> 		 * on x86 sometime uses locked ops because of PPro
> 		 * errata 66, 92, so unless somebody can guarantee
> 		 * atomic_set() here would be safe on all archs (and
> 		 * not only on x86), it's safer to use atomic_add().
> 		 */
> 		atomic_add(page_mapcount(page) + page_mapcount(page_tail) + 1,
> 			   &page_tail->_count);

We don't have anything like get_page_unless_zero() for _mapcount as far as
I can see. And we have similar assignment there now:

	page_tail->_mapcount = page->_mapcount;

Anyway the assignment goes away by the end of patchset.

-- 
 Kirill A. Shutemov