From mboxrd@z Thu Jan 1 00:00:00 1970 From: Hugh Dickins Subject: Re: [PATCH v16 00/22] per memcg lru_lock Date: Sun, 19 Jul 2020 21:47:49 -0700 (PDT) Message-ID: References: <1594429136-20002-1-git-send-email-alex.shi@linux.alibaba.com> <57c619e7-da7e-198e-3de8-530bf19b9450@linux.alibaba.com> <5f2401d3-dd4f-cbc6-8cb4-4e92fc64998c@linux.alibaba.com> Mime-Version: 1.0 Content-Type: MULTIPART/MIXED; BOUNDARY="0-1916823428-1595220486=:1478" Return-path: DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20161025; h=date:from:to:cc:subject:in-reply-to:message-id:references :user-agent:mime-version; bh=6Nps254wO3Vku/AMs/COcTPJnAHzxFnu9/tet83Jczo=; b=Orx+a9RBBcSgVIEOHFVB374CoAHh0H+dcciHwze7bjd/evrRLoNHoEeHX6AUUXAWPE rlEZJuyYrLtfrbxBEr1f6tOk1hUCeuxitWVtbHQSm1gD5eBMeQnnQmiecOui/n+Wy+CH lVR+yKJxXOVRAwcOkkKdPQDX9l0WSRYT55r6M1O9qVGNeakNm5eWCBq/poyfEr+Mh+xO fvXDBer896r9lWpDPRKLzlOfw7HMuYx8YhXqqDQ9X/jiqcowwnnertv+u6tOMyTf4L6x VhxLTL6Tuzj5Wcb/bIoRfQ7SJeG3OQOLcWf9uL2ccqf9yxAULz7Zp0JITwEJl9p0MruI UiVg== In-Reply-To: <5f2401d3-dd4f-cbc6-8cb4-4e92fc64998c-KPsoFbNs7GizrGE5bRqYAgC/G2K4zDHf@public.gmane.org> Sender: cgroups-owner-u79uwXL29TY76Z2rM5mHXA@public.gmane.org List-ID: To: Alex Shi Cc: Hugh Dickins , Alexander Duyck , Andrew Morton , Mel Gorman , Tejun Heo , Konstantin Khlebnikov , Daniel Jordan , Yang Shi , Matthew Wilcox , Johannes Weiner , kbuild test robot , linux-mm , LKML , cgroups-u79uwXL29TY76Z2rM5mHXA@public.gmane.org, Shakeel Butt , Joonsoo Kim , Wei Yang , "Kirill A. Shutemov" This message is in MIME format. The first part should be readable text, while the remaining parts are likely unreadable without MIME-aware tools. --0-1916823428-1595220486=:1478 Content-Type: TEXT/PLAIN; charset=UTF-8 Content-Transfer-Encoding: QUOTED-PRINTABLE On Mon, 20 Jul 2020, Alex Shi wrote: > =E5=9C=A8 2020/7/19 =E4=B8=8B=E5=8D=8811:23, Hugh Dickins =E5=86=99=E9=81= =93: > > I noticed that 5.8-rc5, with lrulock v16 applied, took significantly > > longer to run loads than without it applied, when there should have bee= n > > only slight differences in system time. Comparing /proc/vmstat, somethi= ng > > that stood out was "pgrotated 0" for the patched kernels, which led her= e: > >=20 > > If pagevec_lru_move_fn() is now to TestClearPageLRU (I have still not > > decided whether that's good or not, but assume here that it is good), > > then functions called though it must be changed not to expect PageLRU! > >=20 > > Signed-off-by: Hugh Dickins >=20 > Good catch! >=20 > Thanks a lot, Hugh!=20 > except 6 changes should apply, looks we add one more in swap.c file to st= op > !PageRLU further actions! Agreed, that's a minor optimization that wasn't done before, that can be added (but it's not a fix like the rest of them). >=20 > Many Thanks! > Alex >=20 > @@ -649,7 +647,7 @@ void deactivate_file_page(struct page *page) > * In a workload with many unevictable page such as mprotect, > * unevictable page deactivation for accelerating reclaim is poin= tless. > */ > - if (PageUnevictable(page)) > + if (PageUnevictable(page) || !PageLRU(page)) > return; >=20 > if (likely(get_page_unless_zero(page))) { --0-1916823428-1595220486=:1478--