From mboxrd@z Thu Jan 1 00:00:00 1970 From: =?UTF-8?B?67Cw7ISd7KeE?= Subject: RE:(2) [Kernel][NET] Bug report on packet defragmenting Date: Thu, 08 Nov 2018 11:05:23 +0900 Message-ID: <20181108020523epcms1p55a0c28d3e881a079231fe813258602f6@epcms1p5> References: <91b43bec-cb19-b94b-8ee3-26979e3a19d1@gmail.com> <20181108012927epcms1p47f719c1908da64a378690362901644ee@epcms1p4> Reply-To: soukjin.bae@samsung.com Mime-Version: 1.0 Content-Type: text/plain; charset="utf-8" Content-Transfer-Encoding: quoted-printable To: Eric Dumazet , "netdev@vger.kernel.org" Return-path: Received: from mailout3.samsung.com ([203.254.224.33]:59228 "EHLO mailout3.samsung.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1728379AbeKHLif (ORCPT ); Thu, 8 Nov 2018 06:38:35 -0500 Received: from epcas1p2.samsung.com (unknown [182.195.41.46]) by mailout3.samsung.com (KnoxPortal) with ESMTP id 20181108020526epoutp036a4b6dc6963a575d8236525ff2b59e68~lA_REjcwJ1732317323epoutp03O for ; Thu, 8 Nov 2018 02:05:26 +0000 (GMT) In-Reply-To: <91b43bec-cb19-b94b-8ee3-26979e3a19d1@gmail.com> Sender: netdev-owner@vger.kernel.org List-ID: --------- Original Message --------- Sender : Eric Dumazet Date : 2018-11-08 10:44 (GMT+9) Title : Re: =5BKernel=5D=5BNET=5D Bug report on packet defragmenting =20 > On 11/07/2018 05:29 PM, =EB=B0=B0=EC=84=9D=EC=A7=84=20wrote:=0D=0A>=20=20= =0D=0A>=20>=20If=20ipv6_defrag=20hook=20is=20not=20excuted=20simultaneously= ,=20then=20it's=20ok.=0D=0A>=20>=20ipv6_defrag=20hook=20can=20handle=20that= .=20=5Bexam=203=5D=0D=0A>=20=20=0D=0A>=20This=20seems=20wrong.=0D=0A>=20=20= =0D=0A>=20This=20is=20the=20root=20cause,=20we=20should=20not=20try=20to=20= work=20around=20it=20but=20fix=20it.=0D=0A>=20=20=0D=0A>=20There=20is=20no= =20guarantee=20that=20RSS/RPS/RFS=20can=20help=20here,=20packets=20can=20si= t=20in=20per-cpu=0D=0A>=20backlogs=20long=20enough=20to=20reproduce=20the= =20issue,=20if=20RX=20queues=20interrupts=20are=20spread=0D=0A>=20over=20ma= ny=20cpus.=0D=0A=20=0D=0A=0D=0ADear=20Dumazet,=0D=0A=0D=0AEven=20if=20rx=20= irq=20be=20spread=20to=20overal=20cpu,=20hash=20will=20be=20made=20by=20src= /des=20address.=0D=0Athen=20they'll=20have=20a=20same=20hash=20and=20cpu.= =20is=20it=20not=20enough?=0D=0ADid=20you=20mean=20that=20we=20need=20total= =20solution=20for=20all=20steering=20method?=20not=20just=20only=20RPS?=0D= =0A=0D=0ABest=20regards.=0D=0A=0D=0A