From mboxrd@z Thu Jan 1 00:00:00 1970 From: Thomas Monjalon Subject: Re: [PATCH v2 1/2] drivers: add common folder Date: Tue, 20 Mar 2018 18:27:53 +0100 Message-ID: <2404641.e51pahgxmm@xps> References: <20180319092726.10153-1-pbhagavatula@caviumnetworks.com> <20180320160624.GB18022@ltp-pvn> <20180320170120.GA4260@bricha3-MOBL.ger.corp.intel.com> Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7Bit Cc: Pavan Nikhilesh , jerin.jacob@caviumnetworks.com, santosh.shukla@caviumnetworks.com, anatoly.burakov@intel.com, lironh@marvell.com, fiona.trahe@intel.com, shreyansh.jain@nxp.com, hemant.agrawal@nxp.com, dev@dpdk.org To: Bruce Richardson Return-path: Received: from out5-smtp.messagingengine.com (out5-smtp.messagingengine.com [66.111.4.29]) by dpdk.org (Postfix) with ESMTP id A94D04CB3 for ; Tue, 20 Mar 2018 18:28:22 +0100 (CET) In-Reply-To: <20180320170120.GA4260@bricha3-MOBL.ger.corp.intel.com> List-Id: DPDK patches and discussions List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Errors-To: dev-bounces@dpdk.org Sender: "dev" 20/03/2018 18:01, Bruce Richardson: > On Tue, Mar 20, 2018 at 09:36:25PM +0530, Pavan Nikhilesh wrote: > > Hi Thomas, > > > > On Tue, Mar 20, 2018 at 04:45:47PM +0100, Thomas Monjalon wrote: > > > 20/03/2018 15:40, Pavan Nikhilesh: > > > > Add driver/common folder and skeleton makefile for adding commonly > > > > used functions across mempool, event and net devices. > > > > > > I am not sure there is a real benefit of introducing such new > > > directory. Is it only solving the choice of where you add files used > > > by several drivers? Is drivers/bus/X/ the directory of choice for > > > platform shared files? > > > > drivers/bus doesn't seem to be a appropriate abstraction common API's > > used across multiple PMD's. > > > > This change is in regards with the discussion on ml: > > http://dpdk.org/ml/archives/dev/2018-March/092822.html > > > > drivers/common seems to be a popular opinion but maybe we could use > > something like drivers/SoC or if you have something in mind please > > suggest. > > > I think drivers/common is a better solution as it's not just platform buses > or SoC's that this occurs with. Even PCI devices can export multiple > functions, such as in the QAT case, which leads to the case where you have > shared code between different driver classes. I don't think the PCI bus > driver is a good place to put QAT shared code. :-) If drivers X and Y share some code, it is possible to link files from X directory in Y library. If we want a more explicit and separate directory for shared code, what would be the name of the sub-directories? drivers/common/qat? drivers/common/octeontx? drivers/common/marvell? drivers/common/mellanox?