[BISECTED] arm64: dts: meson-sm1-odroid-hc4: USB not working

[Neil Armstrong]

On 27/12/2022 11:43, Martin Blumenstingl wrote:
> Hi Pierre-Olivier, Hi Neil,
> 
> On Mon, Dec 26, 2022 at 11:48 PM Pierre-Olivier Mercier
> <nemunaire at nemunai.re> wrote:
>>
>> Hi,
>>
>> On my Odroid-HC4, the USB 2.0 port (the only one available) was not
>> working. While it seems I was not the only one in that case[1], after
>> some experiments, I found that by reverting the following commit:
>>
>> 703e84d6615a4 arm64: dts: meson-sm1-odroid-hc4: disable unused USB PHY0
> Thank you for bisecting the commit!
> 
>> it solves the issue: USB is working on latest stable (6.1.1) with this
>> only patch reverted, while it doesn't with vanilla dtb (with same
>> config otherwise).
> [...]
>> I didn't read the specs, but I also tried to inverse the patch (disable
>> usb2_phy1 intead of usb2_phy0) and it didn't work neither. Apparently
>> usb2_phy0 and usb2_phy1 are both required for some reason...
> After reading this mail yesterday I could not make much sense of it.
> Now, while re-reading it I remembered the time when I worked on GXL
> USB support and wrote the following comment: [0]
>    All USB2 PHYs (which are internally "connected" to the dwc3 roothub)
>    need to be powered on, otherwise USB devices cannot be
>    enumerated (even if just one PHY is disabled and if the device is
>    plugged into another, enabled port)
> 
> This is probably also the reason why I left all USB PHYs enabled by
> default in meson-{gxl,gxm}.dtsi (and so did we in
> meson-g12-common.dtsi). Instead we're disabling the &usb node by
> default (and enabling it for boards that have USB ports).
> For readers not familiar with the default enabled/disabled: typically
> we disable the nodes for IP blocks in the .dtsi and then each
> board.dts enables them on demand.
> 
> Neil, what do you think about this?
> Should Pierre-Olivier send a formal revert (with the problem being
> described as part of the git commit message, such as in [1])?

It's true I haven't tested with only a single USB2 phy enabled, but
since the USB design is a derivative from GXL both USB2 PHYs should be
powered on for USB to work.

Pierre-Olivier, can you send a proper revert patch so I can apply it
to fixes and send it asap ?

Thanks for the report !

Neil

> 
> 
> PS: Neil has probably taken some days off, so please give him more
> than usual time to reply.
> 
> 
> Best regards,
> Martin
> 
> 
> [0] https://lore.kernel.org/all/20170814224542.18257-1-martin.blumenstingl@googlemail.com/T/
> [1] https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=47722220660cfb935e27e62d385959ecc296cddb