From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from mail-pf1-f180.google.com (mail-pf1-f180.google.com [209.85.210.180]) (using TLSv1.2 with cipher ECDHE-RSA-AES128-GCM-SHA256 (128/128 bits)) (No client certificate requested) by smtp.subspace.kernel.org (Postfix) with ESMTPS id 3E7403C1E for ; Fri, 1 Jul 2022 15:47:41 +0000 (UTC) Received: by mail-pf1-f180.google.com with SMTP id n12so2870131pfq.0 for ; Fri, 01 Jul 2022 08:47:41 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20210112; h=mime-version:references:in-reply-to:from:date:message-id:subject:to :cc:content-transfer-encoding; bh=9yv6QUWYDmgQjKt1P93gNeGHmUq3W/So/sIQeZFHIcA=; b=rqzfXCul7tw0OtYOnP55pVCB7g/W9r+ESLzhXTbbeXFyAb7FsPVhGwWf1o6kj4AEw9 yy5yMHnLr97j5o8rrLFjKXtvNiVn9FBMx+OuaStaeI4VtIuzSQQMCYCKdG/m5/+ZvbtJ mgVJXlM7Ih3nxS4V9lsQKs/m0R9tyRektkxlV6RlzNAGVnt03mZ7EX5LY/pDnGeggzJW 5yhjTlYGbhC7Tjeh7iltQpmTBPc0/ob4Ek3moFpXstDBXYsgrX+l1Ee4H94nhyirr8kt PXIyVeGHxqvtHBp9W92BVREWrxfsXLcvqjppHbdeBuO4CTDShzDE2h5XITLVdfm3eTZi 0t7A== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20210112; h=x-gm-message-state:mime-version:references:in-reply-to:from:date :message-id:subject:to:cc:content-transfer-encoding; bh=9yv6QUWYDmgQjKt1P93gNeGHmUq3W/So/sIQeZFHIcA=; b=UhA7VgCaGjzXbvXy7CwXxPyp2gBRXIcn7azDNJRb0v1CM3TflGjCHyLqykZ/xPmJer xhseG94uRXi75kRm80l3Nyjeaw6JNQ2NE1Gu0AOUHn01oC2y0dhflu1d8haGrJpi5+TA rcfIbquPQyWa/GffPOBxUWOPSSQUK6vUmq5j2gU0Pa9t1QAVxqTfMu/W+fV2oYN2SoSs 1DsX03Y1dps51PHnl9rQWVCMwiYm1887RLg2eFvwXIXUZXjSsiHXhHXiZYyFsXZCeTMn eq8y9oiZcBnxo+meHEk8Mm9sZjboprDk/utFgUZjKUWOFJYhAzFDB7IrgvRZ6OsZc2iO MM+Q== X-Gm-Message-State: AJIora9jXxoEhw7qI8UmRagpm+GN3MIoNzWiTJR0xTzuTqXPcvd2d7fS ddzOFqWlVD9KZvenojo4WMVhTv3Js4XiWBl6aEh2Jw== X-Google-Smtp-Source: AGRyM1vA2J1Zo47ff1Z3g/0fDruleRJpKHeRmLkW3pftAi7eIEd7S1S7hhAlQL2NFpy1Yupar9lBK8glaB0jqxoYpPM= X-Received: by 2002:a05:6a00:3307:b0:527:cbdc:d7dc with SMTP id cq7-20020a056a00330700b00527cbdcd7dcmr19540188pfb.85.1656690460501; Fri, 01 Jul 2022 08:47:40 -0700 (PDT) Precedence: bulk X-Mailing-List: mptcp@lists.linux.dev List-Id: List-Subscribe: List-Unsubscribe: MIME-Version: 1.0 References: <20220624070656.GE79500@shbuild999.sh.intel.com> <20220624144358.lqt2ffjdry6p5u4d@google.com> <20220625023642.GA40868@shbuild999.sh.intel.com> <20220627023812.GA29314@shbuild999.sh.intel.com> <20220627123415.GA32052@shbuild999.sh.intel.com> <20220627144822.GA20878@shbuild999.sh.intel.com> <20220628034926.GA69004@shbuild999.sh.intel.com> In-Reply-To: <20220628034926.GA69004@shbuild999.sh.intel.com> From: Shakeel Butt Date: Fri, 1 Jul 2022 08:47:29 -0700 Message-ID: Subject: Re: [net] 4890b686f4: netperf.Throughput_Mbps -69.4% regression To: Feng Tang Cc: Eric Dumazet , Linux MM , Andrew Morton , Roman Gushchin , Michal Hocko , Johannes Weiner , Muchun Song , Jakub Kicinski , Xin Long , Marcelo Ricardo Leitner , kernel test robot , Soheil Hassas Yeganeh , LKML , network dev , linux-s390@vger.kernel.org, MPTCP Upstream , "linux-sctp @ vger . kernel . org" , lkp@lists.01.org, kbuild test robot , Huang Ying , Xing Zhengjun , Yin Fengwei , Ying Xu Content-Type: text/plain; charset="UTF-8" Content-Transfer-Encoding: quoted-printable On Mon, Jun 27, 2022 at 8:49 PM Feng Tang wrote: > > On Mon, Jun 27, 2022 at 06:25:59PM +0200, Eric Dumazet wrote: > > On Mon, Jun 27, 2022 at 4:48 PM Feng Tang wrote: > > > > > > Yes, I also analyzed the perf-profile data, and made some layout chan= ges > > > which could recover the changes from 69% to 40%. > > > > > > 7c80b038d23e1f4c 4890b686f4088c90432149bd6de 332b589c49656a45881bca4e= cc0 > > > ---------------- --------------------------- ------------------------= --- > > > 15722 -69.5% 4792 -40.8% 9300 = netperf.Throughput_Mbps > > > > > > > I simply did the following and got much better results. > > > > But I am not sure if updates to ->usage are really needed that often... > > > > > > diff --git a/include/linux/page_counter.h b/include/linux/page_counter.= h > > index 679591301994d316062f92b275efa2459a8349c9..e267be4ba849760117d9fd0= 41e22c2a44658ab36 > > 100644 > > --- a/include/linux/page_counter.h > > +++ b/include/linux/page_counter.h > > @@ -3,12 +3,15 @@ > > #define _LINUX_PAGE_COUNTER_H > > > > #include > > +#include > > #include > > #include > > > > struct page_counter { > > - atomic_long_t usage; > > - unsigned long min; > > + /* contended cache line. */ > > + atomic_long_t usage ____cacheline_aligned_in_smp; > > + > > + unsigned long min ____cacheline_aligned_in_smp; > > unsigned long low; > > unsigned long high; > > unsigned long max; > > @@ -27,12 +30,6 @@ struct page_counter { > > unsigned long watermark; > > unsigned long failcnt; > > > > - /* > > - * 'parent' is placed here to be far from 'usage' to reduce > > - * cache false sharing, as 'usage' is written mostly while > > - * parent is frequently read for cgroup's hierarchical > > - * counting nature. > > - */ > > struct page_counter *parent; > > }; > > I just tested it, it does perform better (the 4th is with your patch), > some perf-profile data is also listed. > > 7c80b038d23e1f4c 4890b686f4088c90432149bd6de 332b589c49656a45881bca4ecc0= e719635902654380b23ffce908d > ---------------- --------------------------- --------------------------- = --------------------------- > 15722 -69.5% 4792 -40.8% 9300 = -27.9% 11341 netperf.Throughput_Mbps > > 0.00 +0.3 0.26 =C2=B1 5% +0.5 0.51 = +1.3 1.27 =C2=B1 2%pp.self.__sk_mem_raise_allocated > 0.00 +0.3 0.32 =C2=B1 15% +1.7 1.74 = =C2=B1 2% +0.4 0.40 =C2=B1 2% pp.self.propagate_protected_us= age > 0.00 +0.8 0.82 =C2=B1 7% +0.9 0.90 = +0.8 0.84 pp.self.__mod_memcg_state > 0.00 +1.2 1.24 =C2=B1 4% +1.0 1.01 = +1.4 1.44 pp.self.try_charge_memcg > 0.00 +2.1 2.06 +2.1 2.13 = +2.1 2.11 pp.self.page_counter_uncharge > 0.00 +2.1 2.14 =C2=B1 4% +2.7 2.71 = +2.6 2.60 =C2=B1 2% pp.self.page_counter_try_charge > 1.12 =C2=B1 4% +3.1 4.24 +1.1 2.22 = +1.4 2.51 pp.self.native_queued_spin_lock_slowpath > 0.28 =C2=B1 9% +3.8 4.06 =C2=B1 4% +0.2 0= .48 +0.4 0.68 pp.self.sctp_eat_data > 0.00 +8.2 8.23 +0.8 0.83 = +1.3 1.26 pp.self.__sk_mem_reduce_allocated > > And the size of 'mem_cgroup' is increased from 4224 Bytes to 4608. Hi Feng, can you please try two more configurations? Take Eric's patch of adding ____cacheline_aligned_in_smp in page_counter and for first increase MEMCG_CHARGE_BATCH to 64 and for second increase it to 128. Basically batch increases combined with Eric's patch.