Re: [PATCH] Revert "ARM: dts: bcm283x: Fix DTC warnings about missing phy-cells"

[Stefan Wahren]

Hi Arnd,


Am 08.01.2018 um 10:27 schrieb Arnd Bergmann:
> On Mon, Jan 8, 2018 at 10:15 AM, Stefan Wahren <stefan.wahren@xxxxxxxx> wrote:
> > Hi Eric,
> > Am 07.01.2018 um 23:08 schrieb Eric Anholt:
> > > Stefan Wahren <stefan.wahren@xxxxxxxx> writes:
> > >
> > > > This reverts commit 014d6da6cb2525d7f48fb08c705cb130cc7b5f4a.
> > > >
> > > > The DT clean up could trigger an endless deferred probe of DWC2 USB
> > > > driver
> > > > on the Raspberry Pi 2/3. So revert the change until we fixed the probing
> > > > issue.
> > > Why's that?  I found that I needed to enable the generic no-op phy
> > > driver, but other than that it was fine.
> >
> > in order to avoid this regression. Changing the configuration is not a
> > solution for the kernelci guys.
> >
> > Btw
> >
> > CONFIG_NOP_USB_XCEIV=y
> >
> > is already enabled in arm64/defconfig and the issue still occured. Do you
> > mean a different option?
> Obviously we need to fix this, but I really want to understand what exactly
> happened so we can fix the code if possible rather than making the
> dts file incompatible with the binding again.

i fully agree, but dwc2 "hacking" usually requires more time than 
reverting this change.

> Do you have any more insight into how we get into the deferred probe
> situation?

I send this bug report [1] on Friday to linux-usb.

Stefan

[1] - https://marc.info/?l=linux-usb&m=151518314314753&w=2

>         Arnd