Re: [PATCH 15/16] net: phy: adin: add ethtool get_stats support

["Ardelean, Alexandru" <alexandru.Ardelean@analog.com>]

On Tue, 2019-08-06 at 17:46 +0200, Andrew Lunn wrote:
> [External]
> 
> On Tue, Aug 06, 2019 at 07:11:57AM +0000, Ardelean, Alexandru wrote:
> > On Mon, 2019-08-05 at 17:28 +0200, Andrew Lunn wrote:
> > > [External]
> > > 
> > > > +struct adin_hw_stat {
> > > > +	const char *string;
> > > > +static void adin_get_strings(struct phy_device *phydev, u8 *data)
> > > > +{
> > > > +	int i;
> > > > +
> > > > +	for (i = 0; i < ARRAY_SIZE(adin_hw_stats); i++) {
> > > > +		memcpy(data + i * ETH_GSTRING_LEN,
> > > > +		       adin_hw_stats[i].string, ETH_GSTRING_LEN);
> > > 
> > > You define string as a char *. So it will be only as long as it should
> > > be. However memcpy always copies ETH_GSTRING_LEN bytes, doing off the
> > > end of the string and into whatever follows.
> > > 
> > 
> > hmm, will use strlcpy()
> > i blindedly copied memcpy() from some other driver
> 
> Hopefully that driver used const char string[ETH_GSTRING_LEN]. Then a
> memcpy is safe. If not, please let me know what driver you copied.

It was an older Marvell PHY driver (marvell.c) ; in version 4.14.
I used that as an initial work-base for writing the driver.
Then I did the conversion to a newer kernel, then I also had to also consider an older kernel, then I got confused :)

Well, in any case, I am solely considering net-next master (now) for upstreaming this.

> 
> > i'm afraid i don't understand about the snapshot feature you are mentioning;
> > i.e. i don't remember seeing it in other chips;
> 
> It is frequency done at the MAC layer for statistics. You tell the
> hardware to snapshot all the statistics. It atomically makes a copy of
> all the statistics into a set of registers. These values are then
> static, and consistent between counters. You can read them out knowing
> they are not going to change.
> 
> > regarding the danger that stat->reg1 rolls over, i guess that is
> > possible, but it's a bit hard to guard against;
> 
> The normal solution is the read the MSB, the LSB and then the MSB
> again. If the MSB value has changed between the two reads, you know a
> roll over has happened, and you need to do it all again.

hmm; ok
I'll try to look for an existing example for this.

> 
>      Andrew