[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-Id: <20160712.113406.1402587122290887574.davem@davemloft.net>
Date: Tue, 12 Jul 2016 11:34:06 -0700 (PDT)
From: David Miller <davem@...emloft.net>
To: linville@...driver.com
Cc: jorge.garcia.gonzalez@...il.com, jiri@...nulli.us,
ben@...adent.org.uk, netdev@...r.kernel.org
Subject: Re: ethtool TODO list - additional info
From: "John W. Linville" <linville@...driver.com>
Date: Tue, 12 Jul 2016 14:12:52 -0400
> I haven't heard a strong case for why it is necessary.
Stats on large scale setups/systems is problematic and ethtool
is a part of that problem.
Extensibility in general suffers because of the ioctl() interface as
well.
For getting ethtool settings or stats, we have no clean filtering
mechanism.
I could go on and on...
I think before any new features are added to ethtool, we move over to
the netlink based interface and lock the existing ioctl() API in stone.
Powered by blists - more mailing lists