RE: [PATCH net-next 1/6] ethtool: Extend link modes settings uAPI with lanes

From: Danielle Ratson <danieller@nvidia.com>
To: Michal Kubecek <mkubecek@suse.cz>
Cc: Jiri Pirko <jiri@resnulli.us>, Andrew Lunn <andrew@lunn.ch>,
	"Jakub Kicinski" <kuba@kernel.org>,
	Ido Schimmel <idosch@idosch.org>,
	"netdev@vger.kernel.org" <netdev@vger.kernel.org>,
	"davem@davemloft.net" <davem@davemloft.net>,
	Jiri Pirko <jiri@nvidia.com>,
	"f.fainelli@gmail.com" <f.fainelli@gmail.com>,
	mlxsw <mlxsw@nvidia.com>, Ido Schimmel <idosch@nvidia.com>,
	"johannes@sipsolutions.net" <johannes@sipsolutions.net>
Subject: RE: [PATCH net-next 1/6] ethtool: Extend link modes settings uAPI with lanes
Date: Tue, 1 Dec 2020 17:22:44 +0000	[thread overview]
Message-ID: <DM6PR12MB4516576B5D5B225540A1C877D8F40@DM6PR12MB4516.namprd12.prod.outlook.com> (raw)
In-Reply-To: <20201126210748.mzbe7ei3wjhvryym@lion.mk-sys.cz>

> -----Original Message-----
> From: Michal Kubecek <mkubecek@suse.cz>
> Sent: Thursday, November 26, 2020 11:08 PM
> To: Danielle Ratson <danieller@nvidia.com>
> Cc: Jiri Pirko <jiri@resnulli.us>; Andrew Lunn <andrew@lunn.ch>; Jakub Kicinski <kuba@kernel.org>; Ido Schimmel
> <idosch@idosch.org>; netdev@vger.kernel.org; davem@davemloft.net; Jiri Pirko <jiri@nvidia.com>; f.fainelli@gmail.com; mlxsw
> <mlxsw@nvidia.com>; Ido Schimmel <idosch@nvidia.com>; johannes@sipsolutions.net
> Subject: Re: [PATCH net-next 1/6] ethtool: Extend link modes settings uAPI with lanes
> 
> On Wed, Nov 25, 2020 at 10:35:35AM +0000, Danielle Ratson wrote:
> > > > What do you think of passing the link modes you have suggested as
> > > > a bitmask, similar to "supported", that contains only one positive bit?
> > > > Something like that:
> >
> > Hi Michal,
> >
> > Thanks for your response.
> >
> > Actually what I said is not very accurate.
> > In ethtool, for speed 100G and 4 lanes for example, there are few link modes that fits:
> > ETHTOOL_LINK_MODE_100000baseKR4_Full_BIT
> > ETHTOOL_LINK_MODE_100000baseSR4_Full_BIT
> > ETHTOOL_LINK_MODE_100000baseCR4_Full_BIT
> > ETHTOOL_LINK_MODE_100000baseLR4_ER4_Full_BIT
> >
> > The difference is the media. And in the driver we shrink into one bit.
> > But maybe that makes passing a bitmask more sense, or am I missing something?
> 
> But as far as I understand, at any moment, only one of these will be actually in use so that's what the driver should report. Or is the
> problem that the driver cannot identify the media in use? (To be
> precise: by "cannot identify" I mean "it's not possible for the driver to find out", not "current code does not distinguish".)
> 
> Michal

After more investigation, those are my conclusions:
We have two types of supported asics in the driver- one of them is able to distinguish between the medias and the other one doesn't.
So in the first type I can send one bit as you requested from the driver to ethtool but in the other one I can't.

The suggestions I have are:
1. Add a bit that for unknown media for each link (something like ETHTOOL_LINK_MODE_100000unknown_Full_BIT). I am not sure it is even possible or makes sense.
2. Pass the link mode as bitmap.

Do you see any other option?

Thanks.