From mboxrd@z Thu Jan 1 00:00:00 1970 From: Govindarajulu Varadarajan Subject: Re: [PATCH net-next] enic: Add support for adaptive interrupt coalescing Date: Thu, 8 May 2014 11:43:03 +0530 (IST) Message-ID: References: <1399386262-20769-1-git-send-email-_govind@gmx.com> <20140507.170043.896249393177676627.davem@davemloft.net> Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII; format=flowed Cc: _govind@gmx.com, netdev@vger.kernel.org, ssujith@cisco.com, benve@cisco.com, neepatel@cisco.com To: David Miller Return-path: Received: from bgl-iport-3.cisco.com ([72.163.197.27]:13278 "EHLO bgl-iport-3.cisco.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751851AbaEHGXK (ORCPT ); Thu, 8 May 2014 02:23:10 -0400 In-Reply-To: <20140507.170043.896249393177676627.davem@davemloft.net> Sender: netdev-owner@vger.kernel.org List-ID: On Wed, 7 May 2014, David Miller wrote: > From: Govindarajulu Varadarajan <_govind@gmx.com> > Date: Tue, 6 May 2014 19:54:22 +0530 > >> From: Sujith Sankar >> >> Signed-off-by: Sujith Sankar >> Signed-off-by: Govindarajulu Varadarajan <_govind@gmx.com> > > Please never submit patches except absolutely the most extremely > trivial ones without a real commit log message explaining what > the change does, how it does it, and why you implemented it that > way. > Will add more description. Sorry >> +enum hrtimer_restart enic_hrtimer_cb(struct hrtimer *timer) >> +{ >> + return HRTIMER_RESTART; >> +} > > For example, I have no idea why in the world you need an hrtimer whose > function is a complete NOP if the networking hardware does the > interrupt coalescing for you. > The idea was to get time delta in usec, to measure the pkt rate. We want to make damping of coal timer more aggressive. I guess we can use ktime_get() instead of hrtimers. Will resubmit the patch with changes in a week. Let us know if anything else needs to be changed. thanks Govind