On Mon, Sep 14, 2020 at 08:48:10AM -0700, Jakub Kicinski wrote:
> On Mon, 14 Sep 2020 03:48:40 +0200 Andrew Lunn wrote:
> > > static int pause_prepare_data(const struct ethnl_req_info *req_base,
> > > @@ -34,10 +36,17 @@ static int pause_prepare_data(const struct
> > > ethnl_req_info *req_base,
> > >
> > > if (!dev->ethtool_ops->get_pauseparam)
> > > return -EOPNOTSUPP;
> > > +
> > > ret = ethnl_ops_begin(dev);
> > > if (ret < 0)
> > > return ret;
> > > dev->ethtool_ops->get_pauseparam(dev, &data->pauseparam);
> > > + if (req_base->flags & ETHTOOL_FLAG_STATS &&
> > > + dev->ethtool_ops->get_pause_stats) {
> > > + memset(&data->pausestat, 0xff,
> > > + sizeof(struct ethtool_pause_stats));
> >
> > Sorry, i missed v1 of these patches. Maybe this has been commented?
> >
> > Filling with 0xff is odd. I don't know of any other code doing this.
>
> Are you saying it'd be clearer to assign ETHTOOL_STAT_NOT_SET in a loop?
Yes. In the end i figured out this is what you intended. I knew there
had to be more to it than what i was seeing. It would be much more
readable to just set the two values to ETHTOOL_STAT_NOT_SET. And i
doubt it makes any difference to the compile, it is probably rolling
the loop and just doing two assignments anyway.
Andrew
