[PATCH 3/7] ARM: mxs: convert m28evk board to device tree

[Marek Vasut]

Dear Shawn Guo,

> On Mon, Jul 09, 2012 at 09:10:54AM +0200, Marek Vasut wrote:
> > Dear Shawn Guo,
> > 
> > > From: Marek Vasut <marex at denx.de>
> > > 
> > > Signed-off-by: Marek Vasut <marex at denx.de>
> > > Signed-off-by: Shawn Guo <shawn.guo at linaro.org>
> > > ---
> > > 
> > >  arch/arm/boot/dts/imx28.dtsi |   21 ++++
> > >  arch/arm/boot/dts/m28evk.dts |  241
> > > 
> > > ++++++++++++++++++++++++++++++++++++++++++ arch/arm/mach-mxs/mach-mxs.c
> > > |
> > > 
> > >  30 +++++
> > >  3 files changed, 292 insertions(+), 0 deletions(-)
> > >  create mode 100644 arch/arm/boot/dts/m28evk.dts
> > 
> > [...]
> > 
> > > +static void __init m28evk_init(void)
> > > +{
> > > +	update_fec_mac_prop(OUI_DENX);
> > 
> > You need to start the FEC clock here too, otherwise ethernet won't work.
> > Same
> 
> Which clock? "enet_out"?  We enabled it for imx28-evk, but
> I do not know m28evk also needs it, as I haven't seen mach-m28evk.c
> enabling any fec clock.  Instead, I saw the following in the old clock
> code.

Well, this for the old revision of the board.

>         /*
>          * Extra fec clock setting
>          * The DENX M28 uses an external clock source
>          * and the clock output must not be enabled
>          */
>         if (!machine_is_m28evk()) {
>                 reg = __raw_readl(CLKCTRL_BASE_ADDR + HW_CLKCTRL_ENET);
>                 reg &= ~BM_CLKCTRL_ENET_SLEEP;
>                 reg |= BM_CLKCTRL_ENET_CLK_OUT_EN;
>                 __raw_writel(reg, CLKCTRL_BASE_ADDR + HW_CLKCTRL_ENET);
>         }
> 
> Or you meant other fec clock?

I mean these enet_out clock, they need to be enabled.

> > thing as on mx28evk. And I noticed that on other platforms too, so maybe
> > that should be somehow the other way around -- turn on the ethernet
> > clock by default.
> > 
> > The rest seems OK. I'll update the .dts file once we get the rest of the
> > stuff in (SPI and i2c).
> > 
> > btw. is someone converting the mxs-lcd ?
> 
> The lcd should be working right now, with the auxdata used.

Sure, but is someone working on the DT variant?

Best regards,
Marek Vasut