On Thu, Jul 18, 2024 at 04:24, Andrew Lunn <andrew@xxxxxxx> wrote :
> On Wed, Jul 17, 2024 at 09:37:24PM +0200, vtpieter@xxxxxxxxx wrote:
> > From: Pieter Van Trappen <pieter.van.trappen@xxxxxxx>
> >
> > Check the erratum workaround application which ensures in addition
> > that indirect register write and read work as expected.
> >
> > Commit b7fb7729c94f ("net: dsa: microchip: fix register write order in
> > ksz8_ind_write8()") would have been found faster like this.
> >
> > Also fix the register naming as in the datasheet.
>
> We are in the merge window at the moment, so net-next is closed at the
> moment. Please repost in two weeks.
Right sorry I only realize this now. I did some reading up on netdev
patches; will make sure to rebase on net-next tree as well when I
submit the v2 in 2 weeks.
> >
> > /* KSZ87xx Errata DS80000687C.
> > * Module 2: Link drops with some EEE link partners.
> > @@ -1981,8 +1982,13 @@ static int ksz8_handle_global_errata(struct dsa_switch *ds)
> > * KSZ879x/KSZ877x/KSZ876x and some EEE link partners may result in
> > * the link dropping.
> > */
> > - if (dev->info->ksz87xx_eee_link_erratum)
> > - ret = ksz8_ind_write8(dev, TABLE_EEE, REG_IND_EEE_GLOB2_HI, 0);
> > + if (dev->info->ksz87xx_eee_link_erratum) {
> > + ret = ksz8_ind_write8(dev, TABLE_EEE, REG_IND_EEE_GLOB2_LO, 0);
> > + if (!ret)
> > + ret = ksz8_ind_read8(dev, TABLE_EEE, REG_IND_EEE_GLOB2_LO, &data);
> > + if (!ret && data)
> > + dev_err(dev->dev, "failed to disable EEE next page exchange (erratum)\n");
>
> If data is not 0, should it be considered fatal? Maybe return -EIO ?
Indeed that would make more sense; will return -EIO for v2.
Thanks, Pieter
