[U-Boot] [RESEND-2 PATCH 0/4] sunxi: Enable EMAC on A83T boards using Realtek RTL8211E PHY

[Chen-Yu Tsai <wens@kernel.org>]

On Thu, Apr 18, 2019 at 10:07 AM Jagan Teki <jagan@amarulasolutions.com> wrote:
>
> On Thu, Apr 18, 2019 at 10:33 PM Chen-Yu Tsai <wens@kernel.org> wrote:
> >
> > On Thu, Apr 18, 2019 at 9:51 AM Jagan Teki <jagan@amarulasolutions.com> wrote:
> > >
> > > On Thu, Apr 18, 2019 at 10:19 PM Chen-Yu Tsai <wens@kernel.org> wrote:
> > > >
> > > > On Thu, Apr 18, 2019 at 9:42 AM Jagan Teki <jagan@amarulasolutions.com> wrote:
> > > > >
> > > > > On Thu, Apr 18, 2019 at 10:00 PM Chen-Yu Tsai <wens@kernel.org> wrote:
> > > > > >
> > > > > > On Thu, Apr 18, 2019 at 9:15 AM Chen-Yu Tsai <wens@kernel.org> wrote:
> > > > > > >
> > > > > > > On Thu, Apr 18, 2019 at 8:09 AM Chen-Yu Tsai <wens@kernel.org> wrote:
> > > > > > > >
> > > > > > > > On Wed, Apr 17, 2019 at 4:38 AM Jagan Teki <jagan@amarulasolutions.com> wrote:
> > > > > > > > >
> > > > > > > > > Hi,
> > > > > > > > >
> > > > > > > > > On Fri, Apr 12, 2019 at 4:05 PM Chen-Yu Tsai <wens@kernel.org> wrote:
> > > > > > > > > >
> > > > > > > > > > From: Chen-Yu Tsai <wens@csie.org>
> > > > > > > > > >
> > > > > > > > > > (Resending yet again with correct email address now subscribed
> > > > > > > > > >  and with proper cover letter subject. Sorry for the noise.)
> > > > > > > > > >
> > > > > > > > > > Hi everyone,
> > > > > > > > > >
> > > > > > > > > > This series enables EMAC (Ethernet controller) on two A83T boards,
> > > > > > > > > > the Cubietruck Plus and Bananapi M3.
> > > > > > > > > >
> > > > > > > > > > A couple of changes are required to the clock definitions to make the
> > > > > > > > > > compiler happy, as it hasn't been coverted to use the common CLK and
> > > > > > > > > > DM_RESET framework. These changes are not used in the A83T code path.
> > > > > > > > > >
> > > > > > > > > > The other two patches enable the sun8i-emac and Realtek PHY driver in
> > > > > > > > > > their respective defconfigs. The device trees already have the EMAC
> > > > > > > > > > enabled.
> > > > > > > > > >
> > > > > > > > > > Since these are compile time issues, all patches should go through
> > > > > > > > > > the same tree.
> > > > > > > > > >
> > > > > > > > > > Regards
> > > > > > > > > > ChenYu
> > > > > > > > > >
> > > > > > > > > >
> > > > > > > > > > Chen-Yu Tsai (4):
> > > > > > > > > >   sunxi: Fix compilation of sun8i-emac for A83T
> > > > > > > > > >   net: sun8i-emac: Fix compilation for A83T
> > > > > > > > > >   sunxi: Enable EMAC on the Cubietruck Plus
> > > > > > > > > >   sunxi: Enable EMAC on the Bananapi M3
> > > > > > > > >
> > > > > > > > > We have EPHY clock and reset support via respective framework [1]
> > > > > > > > > would you rebase these changes on top this.
> > > > > > > > >
> > > > > > > > > [1] http://git.denx.de/?p=u-boot-sunxi.git;a=commitdiff;h=6b2ccabee2368d059513a9be37c0ffaa4a47ec99
> > > > > > > >
> > > > > > > > Lovely! I'll throw the clock register bit patches out and rework the
> > > > > > > > other stuff.
> > > > > > > >
> > > > > > > > Does anyone else have these boards for testing though? I'm away from
> > > > > > > > my boards for the next two weeks (which is past the merge window).
> > > > > > >
> > > > > > > Looks like the EPHY clock and resets weren't converted. So it still
> > > > > > > needs some sort of fix to build. Do you have anything in the works
> > > > > > > regarding EPHY?
> > > > > >
> > > > > > Looks like sunxi/next has the answer. And the new defconfigs build cleanly.
> > > > > > Would you like to me respin on top with RFT? It doesn't actually matter
> > > > > > what branch the defconfig patches are based on since they apply cleanly to
> > > > > > either branch.
> > > > >
> > > > > Yes, ie what I'm thinking.
> > > > >
> > > > > BTW: the existing code doesn't enable the use_internal_phy for
> > > > > H3_EMAC, any idea? ie what I was thought and prepared this EPHY CLK
> > > > > and RESET support enabled it which indeed enabling use_internal_phy
> > > >
> > > > The original code had:
> > > >
> > > >         if (priv->variant == H3_EMAC) {
> > > >                 int parent = fdt_parent_offset(gd->fdt_blob, offset);
> > > >
> > > >                 if (parent >= 0 &&
> > > >                     !fdt_node_check_compatible(gd->fdt_blob, parent,
> > > >                                 "allwinner,sun8i-h3-mdio-internal"))
> > > >                         priv->use_internal_phy = true;
> > > >         }
> > > >
> > > > in sun8i_emac_eth_ofdata_to_platdata(), which I think is what you're
> > > > asking about?
> > >
> > > Yes, the same but priv->use_internal_phy not assign to true at the end.
> >
> > Doesn't the last line set it to true?
> >
> > I'm not sure what you're asking. Are you asking when you should set
> > priv->use_internal_phy?
>
> Inside if statement seems to failed and which intern not setting
> use_internal_phy to true. so at the end priv->use_internal_phy is
> false (as default) is it the proper behavior, I thought it is wrong.

Not sure why it fails. fdt_node_check_compatible will return 0 if
the node lists the given compatible string. And fdt_parent_offset
returns >= 0 for a valid offset. The logic looks correct:

If the PHY node's parent is compatible with "allwinner,sun8i-h3-mdio-internal",
set priv->use_internal_phy = true.

ChenYu