From mboxrd@z Thu Jan 1 00:00:00 1970 From: Florian Fainelli Subject: Re: [RFC 1/2] net/ethtool: Add new coalescing parameter for queue Date: Tue, 08 Dec 2015 10:15:53 -0800 Message-ID: <56671E59.40402@gmail.com> References: <1449549750-12448-1-git-send-email-kan.liang@intel.com> Mime-Version: 1.0 Content-Type: text/plain; charset=utf-8 Content-Transfer-Encoding: 7bit Cc: jesse.brandeburg@intel.com, andi@firstfloor.org, alexander.duyck@gmail.com, jeffrey.t.kirsher@intel.com, shannon.nelson@intel.com, carolyn.wyborny@intel.com, donald.c.skidmore@intel.com, mitch.a.williams@intel.com, ogerlitz@mellanox.com, edumazet@google.com, jiri@mellanox.com, sfeldma@gmail.com, gospo@cumulusnetworks.com, sasha.levin@oracle.com, dsahern@gmail.com, tj@kernel.org, cascardo@redhat.com, corbet@lwn.net To: kan.liang@intel.com, netdev@vger.kernel.org, davem@davemloft.net, bwh@kernel.org Return-path: Received: from mail-pa0-f48.google.com ([209.85.220.48]:33838 "EHLO mail-pa0-f48.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751703AbbLHSQD (ORCPT ); Tue, 8 Dec 2015 13:16:03 -0500 Received: by pacwq6 with SMTP id wq6so15602292pac.1 for ; Tue, 08 Dec 2015 10:16:02 -0800 (PST) In-Reply-To: <1449549750-12448-1-git-send-email-kan.liang@intel.com> Sender: netdev-owner@vger.kernel.org List-ID: On 07/12/15 20:42, kan.liang@intel.com wrote: > From: Kan Liang > > Intrdouce "queue" option for coalesce getting and setting. > For coalesce getting, only the coalescing parameters from specific > queue will be passed to user space. > For coalesce setting, the coalescing parameters will only be applied to > specific queue. > If the queue is set to -1, the coalescing parameters will apply to all > queues. This looks like a good start, but there are a few things that need to be clarified, in particular: - if the number of TX and RX queues differ, but the ethtool coalesce structure contains parameters that affect both the RX and TX side, and the queue number is invalid/non-existent for one of these sides, what is the expected outcome? Same question with specifying a queue number, with RX queue N not belonging to the same queue pair as TX queue N? - from an user perspective do we want to iterate over all queues to set their parameters, or should we have a queue bitmask parameter which allows setting them with the same settings in one shot? What would be the appropriate bitmask size then (32-bits with 16-bits for TX and 16-bits for RX might be too small)? Thanks -- Florian