Re: [PATCH net-next v2] r8169: use the extended tally counter available from RTL8125

[Simon Horman <horms@kernel.org>]

On Thu, Oct 10, 2024 at 10:08:02PM +0200, Heiner Kallweit wrote:
> On 10.10.2024 12:50, Heiner Kallweit wrote:
> > The new hw stat fields partially duplicate existing fields, but with a
> > larger field size now. Use these new fields to reduce the risk of
> > overflows. In addition add support for relevant new fields which are
> > available from RTL8125 only.
> > 
> > Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>

...

> > @@ -1873,13 +1888,33 @@ static void rtl8169_get_ethtool_stats(struct net_device *dev,
> >  	data[10] = le32_to_cpu(counters->rx_multicast);
> >  	data[11] = le16_to_cpu(counters->tx_aborted);
> >  	data[12] = le16_to_cpu(counters->tx_underrun);
> > +
> > +	if (rtl_is_8125(tp)) {
> > +		data[5] = le32_to_cpu(counters->align_errors32);
> > +		data[10] = le64_to_cpu(counters->rx_multicast64);
> > +		data[11] = le32_to_cpu(counters->tx_aborted32);
> > +		data[12] = le32_to_cpu(counters->tx_underrun32);
> > +
> > +		data[13] = le64_to_cpu(counters->tx_octets);
> > +		data[14] = le64_to_cpu(counters->rx_octets);
> > +		data[15] = le32_to_cpu(counters->tx_pause_on);
> > +		data[16] = le32_to_cpu(counters->tx_pause_off);
> > +		data[17] = le32_to_cpu(counters->rx_pause_on);
> > +		data[18] = le32_to_cpu(counters->rx_pause_off);
> > +	}
> >  }
> >  
> >  static void rtl8169_get_strings(struct net_device *dev, u32 stringset, u8 *data)
> >  {
> >  	switch(stringset) {
> >  	case ETH_SS_STATS:
> > +		struct rtl8169_private *tp = netdev_priv(dev);
> 
> patchwork lists the following warning for a clang build:
> warning: label followed by a declaration is a C23 extension [-Wc23-extensions]
> 
> gcc 14.2.1 however had no problem with this code, and also checkpatch didn't
> complain. Is this code acceptable or should it be changed?

Yes, I see that too.
My feeling is that it would be best to change it as it
seems likely that this will break compilation somewhere.