From mboxrd@z Thu Jan 1 00:00:00 1970 From: Eric Dumazet Subject: Re: [PATCH] net: less interrupt masking in NAPI Date: Wed, 03 Dec 2014 18:51:45 -0800 Message-ID: <1417661505.22424.1.camel@edumazet-glaptop2.roam.corp.google.com> References: <1414937973.31792.37.camel@edumazet-glaptop2.roam.corp.google.com> <20141103.122538.387451917276174830.davem@davemloft.net> <547EBC66.4040301@huawei.com> <1417592482.5303.132.camel@edumazet-glaptop2.roam.corp.google.com> <547ED728.2010703@huawei.com> <1417607546.5303.133.camel@edumazet-glaptop2.roam.corp.google.com> <547FC2A1.60105@huawei.com> Mime-Version: 1.0 Content-Type: text/plain; charset="UTF-8" Content-Transfer-Encoding: 7bit Cc: David Miller , netdev@vger.kernel.org, willemb@google.com To: Yang Yingliang Return-path: Received: from mail-ie0-f176.google.com ([209.85.223.176]:58599 "EHLO mail-ie0-f176.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751349AbaLDCvy (ORCPT ); Wed, 3 Dec 2014 21:51:54 -0500 Received: by mail-ie0-f176.google.com with SMTP id tr6so15002812ieb.21 for ; Wed, 03 Dec 2014 18:51:53 -0800 (PST) In-Reply-To: <547FC2A1.60105@huawei.com> Sender: netdev-owner@vger.kernel.org List-ID: On Thu, 2014-12-04 at 10:10 +0800, Yang Yingliang wrote: > On 2014/12/3 19:52, Eric Dumazet wrote: > > On Wed, 2014-12-03 at 17:26 +0800, Yang Yingliang wrote: > > > >> Before this patch, when a large network flow arrives, some other processes > >> response slowly or even don't response because the cpu is dealing with softirq. > >> > >> After this patch, under pressure, much more softirq is doing in ksoftirqd. The other > >> processes be scheduled. > >> > >> My system has dual core. > > > > Which NIC driver are you using ? > > > The driver we developed ourself, it's not upstreamed. Is it a NAPI driver ? Quite frankly this patch is not a stable candidate, as it was a performance improvement, with some regressions in a couple of drivers.