Re: Aw: Re: Re: [PATCH] arm64: dts: rockchip: Fix clocks for rk356x usb

From: Peter Geis
Date: Sat Apr 09 2022 - 07:14:24 EST


On Sat, Apr 9, 2022 at 7:01 AM Heiko Stuebner <heiko@xxxxxxxxx> wrote:
>
> Am Samstag, 9. April 2022, 12:57:39 CEST schrieb Frank Wunderlich:
> > Hi
> > > Gesendet: Samstag, 09. April 2022 um 12:40 Uhr
> > > Von: "Dan Johansen" <strit@xxxxxxxxxxx>
> >
> > > So the issue is only with usb 3 ports, not usb 2 ports?
> >
> > my board has no standalone usb2-ports. usb2 is integrated into the usb3 ports (dual phy). here both were not working.
> >
> > afaik rk3566 has standalone usb2 ports that may not be broken, but i have no such board for testing.

Good Morning,

>
> As far as I understand the issue now after checking the code, this
> patch actually fixes the usb3 series from Peter, right?
>
> I.e. the usb-nodes that are fixed in this patch are not yet present
> in the main rk356x dtsi and only get added in
> "arm64: dts: rockchip: add rk356x dwc3 usb3 nodes" [0]
>
> As we don't want to add broken changes, this fix should squashed
> into a next version of the patch adding the nodes.

Thank you for reporting this, I will squash this fix in and add your signed-off.

However the offending patch is in fact the clock separation patch, and
it breaks backwards compatibility with the rk3328 dtsi which is why my
series also is broken.

The rockchip,dwc3.yaml needs to be fixed to align with the
snps,dwc3.yaml, and both the rk3328 and rk3399 clock names updated.
Also the offending clock separation patch needs a fix to grab the old
clock names for rk3328 backwards compatibility to be retained.

This might also be a good time to look into moving rk3399 to the core
dwc3 driver?

This is a delightful mess.

>
>
> Heiko
>
> [0] https://lore.kernel.org/r/20220408151237.3165046-4-pgwipeout@xxxxxxxxx
>
>

Very Respectfully,
Peter Geis