On 9/28/2020 4:39 PM, Jakub Kicinski wrote:

On Mon, 28 Sep 2020 15:36:50 -0700 Florian Fainelli wrote:

quoted

On 9/28/2020 3:35 PM, Jakub Kicinski wrote:

quoted

On Tue, 29 Sep 2020 00:07:30 +0200 Andrew Lunn wrote:

quoted

On Mon, Sep 28, 2020 at 10:05:08PM +0000, Vladimir Oltean wrote:

quoted

On Mon, Sep 28, 2020 at 02:31:55PM -0700, Jakub Kicinski wrote:

quoted

On Sat, 26 Sep 2020 23:06:26 +0200 Andrew Lunn wrote:

quoted

Not all ports of a switch need to be used, particularly in embedded
systems. Add a port flavour for ports which physically exist in the
switch, but are not connected to the front panel etc, and so are
unused.

This is missing the explanation of why reporting such ports makes sense.

Because this is a core devlink patch, we're talking really generalistic
here.

Hi Vladimir

I don't think Jakub is questioning the why. He just wants it in the
commit message.

Ack, I think we need to clearly say when those should be exposed.
Most ASICs will have disabled ports, and we don't expect NICs to
suddenly start reporting ports for all PCI PFs they may have.

Also I keep thinking that these ports and all their objects should
be hidden under some switch from user space perspective because they
are unlikely to be valuable to see for a normal user. Thoughts?

Hidden in what sense? They are already hidden in that there is no
net_device object being created for them. Are you asking for adding
another option to say, devlink show like:

devlink show -a

which would also show the ports that are disabled during a dump?

Yup, exactly. Looks like ip uses -a for something I don't quite
understand - but some switch along those lines. We already have
-d for hiding less-relevant attributes.

Do you think this is an overkill? I don't feel strongly.

That makes sense to me as it would be confusing to suddenly show unused 
port flavors after this patch series land. Andrew, Vladimir, does that 
work for you as well?
-- 
Florian