Re: [PATCH v3 3/3] arm64: dts: meson: Add minimal support for Odroid-N2

[Neil Armstrong]

On 21/05/2019 20:01, Martin Blumenstingl wrote:
> Hi Neil,
> 
> On Tue, May 21, 2019 at 5:20 PM Neil Armstrong <narmstrong@xxxxxxxxxxxx> wrote:
> [...]
>> +       hub_5v: regulator-hub_5v {
>> +               compatible = "regulator-fixed";
>> +               regulator-name = "HUB_5V";
>> +               regulator-min-microvolt = <5000000>;
>> +               regulator-max-microvolt = <5000000>;
>> +               vin-supply = <&vcc_5v>;
>> +
>> +               gpio = <&gpio GPIOH_5 GPIO_ACTIVE_HIGH>;
> I missed this in the review of v1:
> according to the schematics GPIOH_5 is routed to GL3523 (soldered down
> USB hub) CHIP_EN signal.
> The datasheet [0] mentions that this will "Disable whole chip and keep
> hub in lowest power state (standby mode)"

In the v1, we didn't have the schematics, so it was all a wild guess.

> 
> do you know if this is a similar case as GPIOH_4 (USB hub reset line,
> we configure this using a gpio-hog)?


We have 3 signals for USB :
- GPIOH_5: goes to CHIP_EN to CHIPENABLE to the Hub
- GPIOH_4: goes to RST_N to RESETJ to the Hub
- GPIOH_6: goes to PWREN for the OTG connector PWREN

As you said, the datasheet mentions : "
Disable whole chip and keep hub in lowest power state (standby mode
"

I modeled it as a regulator since when enabled it actually enables the
power on the USB3 ports using the post-Hub VBUSCTRL signal.

Since I can't model these post-hub power enable signals, I can only
model the hub CHIPENABLE as a regulator to handle power.
I can't model the hub as a device subnode inside the dwc3 since
when CHIPENABLE is low, it won't appear...

We could model it as a Hog, but we won't be able to power down
the usb3 ports.

> 
> [...]
>> +&ext_mdio {
>> +       external_phy: ethernet-phy@0 {
>> +               /* Realtek RTL8211F (0x001cc916) */
>> +               reg = <0>;
>> +               max-speed = <1000>;
>> +               eee-broken-1000t;
> are we in the same situation that we have on the X96 Max where network
> dies without eee-broken-1000t?

I assume, let me run more tests to be sure.

> 
> [...]
>> +&usb2_phy0 {
>> +       phy-supply = <&usb_pwr_en>;
> is usb_pwr_en really the phy-supply or is it the vbus-supply of the
> USB top control block (&usb node)?
You're right, now we have the schematics, it's not the phy supply.

I'll move it to vbus-supply since it's connected to the pwrenable
of the otg port.

> 
> if these three questions are answered then you can add my:
> Reviewed-by: Martin Blumenstingl <martin.blumenstingl@xxxxxxxxxxxxxx>

I'll ask you to review the next version before !!!

Thanks.
Neil

> 
> 
> Regards
> Martin
> 
> 
> [0] https://datasheet.lcsc.com/szlcsc/GL3523-OV3S1_C157363.pdf
>