[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20230803120100.2glxdc4yf7sjn7h5@box.shutemov.name>
Date: Thu, 3 Aug 2023 15:01:00 +0300
From: "Kirill A. Shutemov" <kirill.shutemov@...ux.intel.com>
To: Ryan Roberts <ryan.roberts@....com>
Cc: Andrew Morton <akpm@...ux-foundation.org>,
Matthew Wilcox <willy@...radead.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>,
Nathan Chancellor <nathan@...nel.org>,
Alexander Gordeev <agordeev@...ux.ibm.com>,
Gerald Schaefer <gerald.schaefer@...ux.ibm.com>,
linux-kernel@...r.kernel.org, linux-mm@...ck.org
Subject: Re: [PATCH v4 0/3] Optimize large folio interaction with deferred
split
On Wed, Aug 02, 2023 at 05:42:23PM +0100, Ryan Roberts wrote:
> - avoid the split lock contention by using mmu gather (suggested by Kirill)
[Offlist]
So, my idea is to embed struct deferred_split into struct mmu_gather and
make zap path to use it instead of per-node/per-memcg deferred_split. This
would avoid lock contention. If the list is not empty after zap, move the
to the per-node/per-memcg deferred_split.
But it is only relevant if we see lock contention.
--
Kiryl Shutsemau / Kirill A. Shutemov
Powered by blists - more mailing lists