[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <YFstdr1tgRk4jQwJ@lunn.ch>
Date: Wed, 24 Mar 2021 13:15:50 +0100
From: Andrew Lunn <andrew@...n.ch>
To: Moshe Shemesh <moshe@...dia.com>
Cc: Don Bollinger <don@...bollingers.org>,
"'David S. Miller'" <davem@...emloft.net>,
'Jakub Kicinski' <kuba@...nel.org>,
'Adrian Pop' <pop.adrian61@...il.com>,
'Michal Kubecek' <mkubecek@...e.cz>, netdev@...r.kernel.org,
'Vladyslav Tarasiuk' <vladyslavt@...dia.com>
Subject: Re: [RFC PATCH V4 net-next 1/5] ethtool: Allow network drivers to
dump arbitrary EEPROM data
> OK, so following the comments here, I will ignore backward compatibility of
> having global offset and length.
Yes, we can do that in userspace.
> That means I assume in this KAPI I am asked to get data from specific page.
> Should I also require user space to send page number to this KAPI (I mean
> make page number mandatory) ?
It makes the API more explicit. We always need the page, so if it is
not passed you need to default to something. As with addresses, you
have no way to pass back what page was actually read. So yes, make it
mandatory.
And i suppose the next question is, do we make bank mandatory? Once
you have a page > 0x10, you need the bank. Either we need to encode
the logic of when a bank is needed, and make it mandatory then, or it
is always mandatory.
Given the general pattern, we make it mandatory.
But sometime in the future when we get yet another SFP format, with
additional parameters, they will be optional, in order to keep
backwards compatibility.
Andrew
Powered by blists - more mailing lists