Received-SPF: pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
Date:   Thu, 1 Mar 2018 15:39:52 +0100
From:   Martin Schwidefsky <schwidefsky@de.ibm.com>
To:     "Kirill A. Shutemov" <kirill@shutemov.name>
Cc:     Matthew Wilcox <willy@infradead.org>, linux-mm@kvack.org,
        Matthew Wilcox <mawilcox@microsoft.com>,
        linux-kernel@vger.kernel.org
Subject: Re: [PATCH v3 1/4] s390: Use _refcount for pgtables
In-Reply-To: <20180301142855.emaa5x65oj2hkwsm@node.shutemov.name>
References: <20180228223157.9281-1-willy@infradead.org>
        <20180228223157.9281-2-willy@infradead.org>
        <20180301125310.jx6c5dypk5axrmum@node.shutemov.name>
        <20180301150420.19a14fd3@mschwideX1>
        <20180301142855.emaa5x65oj2hkwsm@node.shutemov.name>
MIME-Version: 1.0
Content-Type: text/plain; charset=US-ASCII
Content-Transfer-Encoding: 7bit
Message-Id: <20180301153952.668bfdc7@mschwideX1>
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On Thu, 1 Mar 2018 17:28:55 +0300
"Kirill A. Shutemov" <kirill@shutemov.name> wrote:

> On Thu, Mar 01, 2018 at 03:04:20PM +0100, Martin Schwidefsky wrote:
> > On Thu, 1 Mar 2018 15:53:10 +0300
> > "Kirill A. Shutemov" <kirill@shutemov.name> wrote:
> >   
> > > On Wed, Feb 28, 2018 at 02:31:54PM -0800, Matthew Wilcox wrote:  
> > > > From: Matthew Wilcox <mawilcox@microsoft.com>
> > > > 
> > > > s390 borrows the storage used for _mapcount in struct page in order to
> > > > account whether the bottom or top half is being used for 2kB page
> > > > tables.  I want to use that for something else, so use the top byte of
> > > > _refcount instead of the bottom byte of _mapcount.  _refcount may
> > > > temporarily be incremented by other CPUs that see a stale pointer to
> > > > this page in the page cache, but each CPU can only increment it by one,
> > > > and there are no systems with 2^24 CPUs today, so they will not change
> > > > the upper byte of _refcount.  We do have to be a little careful not to
> > > > lose any of their writes (as they will subsequently decrement the
> > > > counter).    
> > > 
> > > Hm. I'm more worried about false-negative put_page_testzero().
> > > Are you sure it won't lead to leaks. I cannot say from the code changes.
> > > 
> > > And for page-table pages should have planty space in other fields.
> > > IIRC page->mapping is unused there.  
> >  
> > 2^^24 put_page_testzero calls for page table pages? I don't think so.  
> 
> No, I mean oposite: we don't free the page when we should. 2^24 is not
> zero and page won't be freed if the acctual refcount (without the flag in
> upper bits) drops to zero.

Ah, ok. But this is not a problem as the page is freed after both bits for
the two 2K pieces havbe been set to zero.

-- 
blue skies,
   Martin.

"Reality continues to ruin my life." - Calvin.