From mboxrd@z Thu Jan 1 00:00:00 1970 From: Shahaf Shuler Subject: Re: [PATCH 00/13] cxgbe: add CXGBE VF PMD and updates Date: Wed, 28 Mar 2018 08:30:58 +0000 Message-ID: References: <20180327070155.GA17402@chelsio.com> <20180328073933.GA19600@chelsio.com> Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: quoted-printable Cc: Ferruh Yigit , "dev@dpdk.org" , Kumar A S , Nirranjan Kirubaharan , Indranil Choudhury To: Rahul Lakkireddy Return-path: Received: from EUR01-HE1-obe.outbound.protection.outlook.com (mail-he1eur01on0064.outbound.protection.outlook.com [104.47.0.64]) by dpdk.org (Postfix) with ESMTP id 6B3581B19 for ; Wed, 28 Mar 2018 10:31:00 +0200 (CEST) In-Reply-To: <20180328073933.GA19600@chelsio.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" Wednesday, March 28, 2018 10:40 AM, Rahul Lakkireddy: > On Wednesday, March 03/28/18, 2018 at 10:19:17 +0530, Shahaf Shuler > > > > What exactly is the offload you try to expose? To strip the inner VLAN = while > keeping the outer one? > > >=20 > Sorry about that, I should have provided more details on this earlier. > Chelsio hardware supports below for double-vlans: > 1) Dropping Outer VLAN in hw; inner vlan is stripped and available as pac= ket > metadata. So this is basically the DEV_RX_OFFLOAD_QINQ_STRIP with limitation from you= r device you cannot return the outer vlan in the packet metadata > 2) Preserving Outer VLAN so, Outer VLAN is stripped and available as pack= et > metadata. Inner VLAN is available as part of packet itself. And this one is the regular DEV_RX_OFFLOAD_VLAN_STRIP >=20 > So, the configurability here is with respect to Outer VLAN and subsequent= ly > what is available as part of packet metadata (outer or inner vlan). Inne= r > VLAN gets retained in both cases. However, based on how Outer VLAN is > configured, Inner VLAN will be available either as packet metadata or as = part > of packet itself. >=20 > We are not very sure if DEV_RX_OFFLOAD_QINQ_STRIP could be used here. > If it holds the same meaning then we will use the same. Yeah , we are missing good documentation of what each offload means.=20 >=20 > Thanks, > Rahul