Re: [PATCH v8 1/6] software node: rename is_array to is_inline

[Rafael J. Wysocki]

On Fri, Dec 13, 2019 at 7:47 AM Marek Szyprowski
<m.szyprowski@xxxxxxxxxxx> wrote:
>
> Hi Rafael,
>
> On 12.12.2019 17:41, Rafael J. Wysocki wrote:
> > On Thu, Dec 12, 2019 at 12:28 PM Andy Shevchenko
> > <andriy.shevchenko@xxxxxxxxxxxxxxx> wrote:
> >> On Thu, Dec 12, 2019 at 12:12:36PM +0100, Marek Szyprowski wrote:
> >>> On 08.11.2019 05:22, Dmitry Torokhov wrote:
> >>>> We do not need a special flag to know if we are dealing with an array,
> >>>> as we can get that data from ratio between element length and the data
> >>>> size, however we do need a flag to know whether the data is stored
> >>>> directly inside property_entry or separately.
> >>>>
> >>>> Signed-off-by: Dmitry Torokhov <dmitry.torokhov@xxxxxxxxx>
> >>> Today I've noticed that this patch got merged to linux-next as commit
> >>> e6bff4665c595b5a4aff173848851ed49ac3bfad. Sadly it breaks DWC3/xHCI
> >>> driver operation on Samsung Exynos5 SoCs (and probably on other SoCs
> >>> which use DWC3 in host mode too). I get the following errors during boot:
> >>>
> >>> dwc3 12000000.dwc3: failed to add properties to xHCI
> >>> dwc3 12000000.dwc3: failed to initialize host
> >>> dwc3: probe of 12000000.dwc3 failed with error -61
> >>>
> >>> Here is a full kernel log from Exynos5250-based Snow Chromebook on KernelCI:
> >>>
> >>> https://storage.kernelci.org/next/master/next-20191212/arm/exynos_defconfig/gcc-8/lab-collabora/boot-exynos5250-snow.txt
> >>>
> >>> (lack of 'ref' clk is not related nor fatal to the driver operation).
> >>>
> >>> The code which fails after this patch is located in
> >>> drivers/usb/dwc3/host.c. Let me know if I can help more in locating the bug.
> >> Thank you for report.
> >>
> >> I think we should not have that patch in the fist place... I used to have
> >> a bad feeling about it and then forgot about it existence.
> > Well, I think you mean the [2/6].
> >
> > The $subject one really shouldn't change functionality, we must have
> > missed something here.
>
> Nope, I was really talking about [1/6]. It looks that it revealed an
> issue in the DWC3 driver pointed by Dmitry.

Right, but I was referring to the Andy's comment.

Cheers!