From mboxrd@z Thu Jan 1 00:00:00 1970 From: Chas Williams <3chas3@gmail.com> Subject: Re: [PATCH v3 27/27] net/ena: set link speed as none Date: Mon, 11 Jun 2018 12:15:09 -0400 Message-ID: References: <20180607094322.14312-1-mk@semihalf.com> <20180607094322.14312-27-mk@semihalf.com> Mime-Version: 1.0 Content-Type: text/plain; charset="UTF-8" Content-Transfer-Encoding: quoted-printable Cc: Ferruh Yigit , mw@semihalf.com, gtzalik@amazon.com, evgenys@amazon.com, dev@dpdk.org, matua@amazon.com, rk@semihalf.com To: mk@semihalf.com Return-path: Received: from mail-it0-f67.google.com (mail-it0-f67.google.com [209.85.214.67]) by dpdk.org (Postfix) with ESMTP id 7E2AE1DFD5 for ; Mon, 11 Jun 2018 18:15:21 +0200 (CEST) Received: by mail-it0-f67.google.com with SMTP id v83-v6so11610487itc.3 for ; Mon, 11 Jun 2018 09:15:21 -0700 (PDT) In-Reply-To: List-Id: DPDK patches and discussions List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Errors-To: dev-bounces@dpdk.org Sender: "dev" On Mon, Jun 11, 2018 at 4:01 AM Micha=C5=82 Krawczyk wrot= e: > 2018-06-10 3:35 GMT+02:00 Chas Williams <3chas3@gmail.com>: > > > > > > On Fri, Jun 8, 2018 at 3:37 PM Ferruh Yigit > wrote: > >> > >> On 6/7/2018 10:43 AM, Michal Krawczyk wrote: > >> > From: Rafal Kozik > >> > > >> > Link speed should is not limited to 10Gb/s and it shouldn't be > >> > hardcoded. > >> > > >> > They link speed is set to none instead and the applications shouldn'= t > >> > rely on this value when using ENA PMD. > >> > >> Why not able to set link speed? > >> > >> And what is the link_autoneg, ETH_LINK_FIXED? What is the point of > setting > >> link > >> speed FIXED and later speed value NONE? > > > > > > The link speed is part of the spanning tree path computation. I don't > think > > picking 0 (or undefined) > > is the appropriate choice. For virtual interfaces link speed is a > > troublesome quantity but advertising > > undefined seems worse then just use some constant value. > > > > Setting constant value is not making much sense in the virtualized > environment (ENA can be only used in the cloud). Link speed can be > very flexible in that case and that's why ENA isn't providing it. If > we will use constant value, what it should be? It may lead to > confusion or can result in the performance issues because some > applications are relying on this number to send max traffic. > What applications are using this value to determine the transmit speed? > Applications should not make assumption that the link speed is always > giving valid value and if it is not, they should try to configure rate > dynamically, instead of statically. So advertising a fixed speed of 10G should be perfectly fine. The application should not assume 10G is a valid value and will attempt to dynamically configure the rate. The link speed should never be used to determine a transmit rate for an application because you don't know the speeds in the middle of the network. > > >> > >> > >> > > >> > Fixes: 1173fca ("ena: add polling-mode driver") > >> > >> For next time can you please use git alias to have consistent Fixes > format > >> > >> > > >> > Signed-off-by: Rafal Kozik > >> > Acked-by: Michal Krawczyk > >> > --- > >> > drivers/net/ena/ena_ethdev.c | 2 +- > >> > 1 file changed, 1 insertion(+), 1 deletion(-) > >> > > >> > diff --git a/drivers/net/ena/ena_ethdev.c > b/drivers/net/ena/ena_ethdev.c > >> > index 5c3b6494f..9ae73e331 100644 > >> > --- a/drivers/net/ena/ena_ethdev.c > >> > +++ b/drivers/net/ena/ena_ethdev.c > >> > @@ -848,7 +848,7 @@ static int ena_link_update(struct rte_eth_dev > *dev, > >> > adapter =3D (struct ena_adapter *)(dev->data->dev_private); > >> > > >> > link->link_status =3D adapter->link_status ? ETH_LINK_UP : > >> > ETH_LINK_DOWN; > >> > - link->link_speed =3D ETH_SPEED_NUM_10G; > >> > + link->link_speed =3D ETH_SPEED_NUM_NONE; > >> > link->link_duplex =3D ETH_LINK_FULL_DUPLEX; > >> > > >> > return 0; > >> > > >> > > >