Re: [PATCH net-next v2 7/9] eth: bnxt: maintain rx pkt/byte stats in SW

[Jakub Kicinski <kuba@kernel.org>]

On Sun, 2 Mar 2025 23:06:29 -0800 Michael Chan wrote:
> On Thu, Feb 27, 2025 at 5:25 PM Jakub Kicinski <kuba@kernel.org> wrote:
> >
> > Some workloads want to be able to track bandwidth utilization on
> > the scale of 10s of msecs. Updating all HW stats at this rate is
> > both hard and wasteful of PCIe bandwidth.
> >
> > Maintain basic Rx pkt/byte counters in software. ethtool -S will still
> > show the HW stats, but qstats and rtnl stats will show SW statistics.
> >
> > We need to take care of HW-GRO, XDP and VF representors. Per netdev
> > qstat definition Rx stats should reflect packets passed to XDP (if
> > active, otherwise to the stack). XDP and GRO do not interoperate
> > in bnxt, so we need to count the packets in a few places.
> > Add a helper and call it where needed.
> >
> > Do not count VF representor traffic as traffic for the main netdev.
> >
> > The stats are added towards the end of the struct since ethtool
> > code expects existing members to be first.
> >
> > Signed-off-by: Jakub Kicinski <kuba@kernel.org>  
> 
> The byte counting looks correct except for VLAN packets with the VLAN
> tag stripped.  Do we want to count the 4 bytes for the VLAN tag?

Not 100% clear to me. I looked at nfp and a couple of Intel drivers
and if I read the code right they don't count VLAN tags if stripped.
I suspect that's what most drivers which count in SW will do, simply
because it's easier rather than intentional choice by their authors.

Happy to change if you have a preference, but to spell it out my
understanding is that not counting stripped tags was more comment 
and that's what I intended to implement here.