From mboxrd@z Thu Jan 1 00:00:00 1970 From: "Joseph, Anoob" Subject: Re: [PATCH v1 00/15] preparing l2fwd for eventmode additions Date: Tue, 3 Jul 2018 18:46:39 +0530 Message-ID: <1390dd35-cd67-a67f-35b5-1d328f82a13a@caviumnetworks.com> References: <1528477766-15788-1-git-send-email-anoob.joseph@caviumnetworks.com> <1528976946-14396-1-git-send-email-anoob.joseph@caviumnetworks.com> <20180619100939.GA23348@bricha3-MOBL.ger.corp.intel.com> <827adfdd-602f-a70c-fe9f-2c821ceaafb3@caviumnetworks.com> Mime-Version: 1.0 Content-Type: text/plain; charset=windows-1252; format=flowed Content-Transfer-Encoding: 7bit Cc: Pablo de Lara , "Jacob, Jerin" , "Athreya, Narayana Prasad" , "dev@dpdk.org" , Hemant Agrawal , Nikhil Rao , Sunil Kumar Kori , gage.eads@intel.com, harry.van.haaren@intel.com, narender.vangati@intel.com, "Bhagavatula, Pavan" , Thomas Monjalon , Akhil Goyal To: Bruce Richardson Return-path: Received: from NAM03-DM3-obe.outbound.protection.outlook.com (mail-dm3nam03on0086.outbound.protection.outlook.com [104.47.41.86]) by dpdk.org (Postfix) with ESMTP id 05D501BE59 for ; Tue, 3 Jul 2018 15:15:36 +0200 (CEST) In-Reply-To: <827adfdd-602f-a70c-fe9f-2c821ceaafb3@caviumnetworks.com> Content-Language: en-US List-Id: DPDK patches and discussions List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Errors-To: dev-bounces@dpdk.org Sender: "dev" Hi, Gentle reminder! Thanks, Anoob On 19-06-2018 19:37, Anoob Joseph wrote: > External Email > > Hi Bruce, > > > Thanks for the feedback. Please see inline. > > > + Hemant, Nikhil, Sunil, Gage, Harry, Narender, Pavan, Thomas, Akhil > > > On 19/06/18 15:39, Bruce Richardson wrote: >> On Tue, Jun 19, 2018 at 03:34:29PM +0530, Anoob Joseph wrote: >>> Hi Bruce, Pablo, >>> >>> Any comments on this series? >>> >>> Thanks, >>> Anoob >>> >>> On 14/06/18 17:18, Anoob Joseph wrote: >>>> This patchset modularizes l2fwd application to prepare it for >>>> eventmode >>>> additions. This patchset doesn't change the code flow or logic, except >>>> for few minor improvements. Some of the newly added functions are used >>>> in just one place, but is added for efficient usage with eventmode. >>>> >>>> v1: >>>> * Fix all checkpatch reported issues >>>> >> My main concern here is how much this eventmode addition is going to >> complicate the l2fwd example. l2fwd has always been a pretty basic >> example >> app to get users started on the basics of DPDK use, and I'm not sure how >> much we want to move away from that. Is this eventmode-l2fwd better >> being a >> separate app, to allow l2fwd to be kept as simple as it can be? >> >> Looking for more thoughts from others here, since it's a community >> decision >> as to the scope of the examples. >> >> /Bruce > The eventmode helper abstracts most of the changes required by the > application to run in eventmode. This was taken up following the > comments on a patch submitted by Sunil(sunil.kori@nxp.com). > http://patches.dpdk.org/patch/37955/ > > With eventmode helper, an application can be moved to eventmode with > minimal changes. For l2fwd, the key patch which enables eventmode is, > > http://patches.dpdk.org/patch/40920/ > [The aforementioned patch is dependent on this patch series] > > The bulk of the code in this patch(40920) is adding multiple event mode > worker functions.The existing init code and poll mode worker is barely > touched. Multiple workers were introduced because a single event mode > worker would not have made the best use of the varying capabilities of > event devices. > > Single event mode worker could've demonstrated how minimal the changes > can be. But the ability to register multiple workers, fine tuned for > varying capabilities, is a good feature to have since it will enable > applications to utilize the full potential of the hardware. > > Eventmode helper patch series: > http://patches.dpdk.org/project/dpdk/list/?series=61 > > The rules that were followed while drafting eventmode helper were very > simple, > 1. Move any code common to multiple applications to eventmode helper > 2. Expose all capabilities of the devices involved (event & eth devs) > 3. Minimize changes to the existing code > > For l2fwd we can opt for a new eventmode-l2fwd app, but this might not > work for more complicated apps like l3fwd & ipsec-secgw. L2fwd app will > stay the same even with the eventmode additions. It will still be a > quick-start, easy-to-use app. In addition to demonstrating DPDK, it will > also be able to demonstrate how easily an app can be made to run in > eventmode, using the helper functions. > > With more event adapters getting added (tx adapter, crypto adapter, > timer adapter etc), the helper will prove useful in abstracting the > complex configuration options exposed by adapters. Similar changes would > be required in other example apps, and the additions in l2fwd is to > finalize on the approach. > > The current patch series just re-factors the code with couple of patches > fixing preexisting checkpatch issues. The rest of the changes are split > into individual patches for ease of review and testing. Hence the large > number of patches. > > Thanks, > Anoob