[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <ZLVkucGl88vxEuIu@casper.infradead.org>
Date: Mon, 17 Jul 2023 16:56:41 +0100
From: Matthew Wilcox <willy@...radead.org>
To: Ryan Roberts <ryan.roberts@....com>
Cc: Andrew Morton <akpm@...ux-foundation.org>,
Yin Fengwei <fengwei.yin@...el.com>,
David Hildenbrand <david@...hat.com>,
Yu Zhao <yuzhao@...gle.com>, Yang Shi <shy828301@...il.com>,
"Huang, Ying" <ying.huang@...el.com>, Zi Yan <ziy@...dia.com>,
linux-kernel@...r.kernel.org, linux-mm@...ck.org
Subject: Re: [PATCH v1 2/3] mm: Implement folio_remove_rmap_range()
On Mon, Jul 17, 2023 at 04:49:19PM +0100, Ryan Roberts wrote:
> > We're still doing one atomic op per page on the folio's nr_pages_mapped
> > ... is it possible to batch this and use atomic_sub_return_relaxed()?
>
> Good spot, something like this:
>
> } else {
> for (; nr != 0; nr--, page++) {
> /* Is this the page's last map to be removed? */
> last = atomic_add_negative(-1, &page->_mapcount);
> if (last)
> nr_unmapped++;
> }
>
> /* Pages still mapped if folio mapped entirely */
> nr_mapped = atomic_sub_return_relaxed(nr_unmapped, mapped);
> if (nr_mapped >= COMPOUND_MAPPED)
> nr_unmapped = 0;
> }
I think that's right, but my eyes always go slightly crossed trying to
read the new mapcount scheme.
Powered by blists - more mailing lists