Re: [PATCH net-next v1 2/3] net: dsa: ar9331: add support for pause stats

[Vladimir Oltean]

On Mon, Jun 27, 2022 at 09:15:21AM -0700, Jakub Kicinski wrote:
> On Sun, 26 Jun 2022 19:10:08 +0200 Oleksij Rempel wrote:
> > > Is there an authoritative source who is able to tell whether rtnl_link_stats64 ::
> > > rx_packets and tx_packets should count PAUSE frames or not?  
> > 
> > Yes, it will be interesting to know how to proceed with it.
> 
> I'm curious as well, AFAIK most drivers do not count pause to ifc stats.

How do you know? Just because they manually bump stats->tx_bytes and
stats->tx_packets during ndo_start_xmit?

That would be a good assumption, but what if a network driver populates
struct rtnl_link_stats64 entirely based on counters reported by hardware,
including {rx,tx}_{packets,bytes}?

Personally I can't really find a reason why not count pause frames if
you can. And in the same note, why go to the extra lengths of hiding
them as Oleksij does. For example, the ocelot/felix switches do count
PAUSE frames as packets/bytes, both on rx and tx.

> > For example KSZ switch do count pause frame Bytes together will other
> > frames. At same time, atheros switch do not count pause frame bytes
> > at all.
> > 
> > To make things worse, i can manually send pause frame of any size, so
> > it will not be accounted by HW. What ever decision we will made, i
> > will need to calculate typical pause frame size and hope it will fit
> > for 90% of cases.
> 
> :(