On Wed, Aug 7, 2019 at 7:15 PM Fabio Estevam <festevam@xxxxxxxxx> wrote:
>
> Hi Russell,
>
> On Wed, Aug 7, 2019 at 6:13 PM Russell King - ARM Linux admin
> <linux@xxxxxxxxxxxxxxx> wrote:
>
> > Well, the question is what behaviour do we want. The driver's original
> > behaviour prior to it breaking was to ignore DMA if it wasn't found
> > (e.g. when the DMA driver is a module.)
> >
> > Your patch restores that behaviour, which is the safest approach.
>
> Thanks for the feedback.
>
> > Audrey's patch has the effect that none of the I2C buses will be
> > registered if the DMA driver is a module, until the DMA driver is
> > loaded. Since an I2C bus _may_ have the systems RTC on it, not
> > probing the I2C bus will cause the RTC not to be read by the kernel
> > at boot, leading to the system booting with a unixtime of 0.
>
> That's a very good point.
>
> > I haven't checked what the situation is wrt RTCs on the ZII boards,
> > whether they do indeed have external I2C RTCs on them.
>
> Yes, they do have external I2C RTCs. Please check this commit for example:
> 09fc0dacc ("ARM: dts: imx6qdl-zii-rdu2: Disable the internal RTC")
And here is another one for the vf610-zii:
8da0af5d6d39 ("ARM: dts: vf610-zii: Disable SNVS RTC")
> If Wolfram agrees, I think we can go with my patch then.
>
> Thanks
