[PATCH 1/2] arm64: dts: armada-3720-espressobin: sync with downstream

[Pali Rohár]

On Monday 31 August 2020 11:59:28 Andre Heider wrote:
> On 31/08/2020 10:46, Pali Rohár wrote:
> > On Monday 31 August 2020 10:17:07 Andre Heider wrote:
> > > On 31/08/2020 10:01, Pali Rohár wrote:
> > > > On Monday 31 August 2020 05:34:07 Andre Heider wrote:
> > > > > +/* U11 */
> > > > > +&sdhci1 {
> > > > > +	non-removable;
> > > > > +	bus-width = <8>;
> > > > > +	mmc-ddr-1_8v;
> > > > > +	mmc-hs400-1_8v;
> > > > > +	marvell,pad-type = "fixed-1-8v";
> > > > > +	pinctrl-names = "default";
> > > > > +	pinctrl-0 = <&mmc_pins>;
> > > > > +	status = "disabled";
> > > > > +
> > > > > +	#address-cells = <1>;
> > > > > +	#size-cells = <0>;
> > > > > +	mmccard: mmccard at 0 {
> > > > > +		compatible = "mmc-card";
> > > > > +		reg = <0>;
> > > > > +	};
> > > > > +};
> > > > 
> > > > Anyway, Andre, have you tested these Marvell's patches on top of U-Boot
> > > > from master git branch? Because espressobin with eMMC is rather rare (do
> > > > not know who has this board for testing; do you really have it?) and
> > > > from my experience with previous patches I know that Marvell's patches
> > > > does not always work on current upstream U-Boot.
> > > 
> > > I haven't tested it on a espressobin emmc myself, but the node is the same
> > > as for arch/arm/dts/armada-3720-db.dts and almost the same as for
> > > arch/arm/dts/armada-3720-uDPU.dts.
> > 
> > I think changes should be tested prior merged them into upstream.
> > Untested patches can cause other hidden issues.
> 
> Like an utterly broken u-boot for 2 years due to data aborts? :)

Exactly!

> > Also as I wrote, other SD patches for EspressoBin copied+pasted from
> > Marvell's did not worked as is applied on current upstream U-Boot. And
> > also argument that "patch is similar as other node" is not reason to not
> > test changed. As a first thing which I did for SD card support on
> > EspressoBin was that I copied DTS node from Turris MOX... and on
> > EspressoBin it did not worked. On MOX was card working fine.
> 
> I would test the patches if I had the hardware, but I just don't.
> 
> Maybe someone with an espressobin-emmc reads and tests this.
> And I can try to find users on e.g. the openwrt forums.
> But I wouldn't hold my breath...

Another option could be to solder eMMC on existing non-eMMC EspressoBin
board, but this is rather very hard without expensive equipment.

Anyway, if there are no users with hardware I think it does not make
sense to try supporting such HW combination.