Re: [PATCH net-next 2/3] tuntap: allow overriding ethtool driver info

[Stephen Hemminger <stephen@networkplumber.org>]

On Thu, 25 Jul 2013 22:36:22 +0100
Ben Hutchings <bhutchings@solarflare.com> wrote:

> On Thu, 2013-07-25 at 14:32 -0700, Stephen Hemminger wrote:
> > On Thu, 25 Jul 2013 22:19:17 +0100
> > Ben Hutchings <bhutchings@solarflare.com> wrote:
> > 
> > > On Wed, 2013-07-24 at 17:16 -0700, Stephen Hemminger wrote:
> > > > On Thu, 25 Jul 2013 00:48:07 +0100
> > > > Ben Hutchings <bhutchings@solarflare.com> wrote:
> > > > 
> > > > > On Wed, 2013-07-24 at 16:13 -0700, Stephen Hemminger wrote:
> > > > > > This patch adds new ioctl to allow setting the ethtool information
> > > > > > returned by the TUN device. This is useful when using tun device as a surrogate
> > > > > > for hardware or other software emulation.
> > > > > 
> > > > > I don't like this idea.  Which tools are you trying to fool?  How does
> > > > > this work when you don't implement any driver-specific behaviour (e.g.
> > > > > SIOCDEVPRIVATE) they expect?
> > > > 
> > > > We use surrogate interfaces in user mode application and want to
> > > > display different information for these than the normal TUN device.
> > > 
> > > What is the problem that can't be solved purely in userland?
> > > 
> > > Ben.
> > > 
> > 
> > There applications (like SNMP) that use ethtool info.
> 
> Huh, I never expected they would record the driver and version.  So you
> want them to report the userland software name and version?  I can sort
> of see how this might be useful as part of a test harness, but I would
> hate to see this get used in production.  Users ought to be able to
> trust that 'ethtool -i' shows them the kernel driver information.
> 
> Ben.
> 

It is more the bus info.
The user mode driver knows the PCI address of the device it is controlling.