[U-Boot] [PATCHv4 3/3] ARM64: poplar: hi3798cv200: u-boot support for Poplar 96Boards

[Tom Rini]

On Thu, May 25, 2017 at 03:21:04PM -0600, Simon Glass wrote:
> Hi Tom,
> 
> On 25 May 2017 at 15:12, Tom Rini <trini at konsulko.com> wrote:
> > On Thu, May 25, 2017 at 10:58:20PM +0200, Jorge Ramirez wrote:
> >> On 05/25/2017 10:55 PM, Jorge Ramirez wrote:
> >> >On 05/25/2017 10:31 PM, Tom Rini wrote:
> >> >>On Thu, May 25, 2017 at 08:38:47PM +0200, Jorge Ramirez wrote:
> >> >>>On 05/18/2017 12:06 AM, Tom Rini wrote:
> >> >>>>>>>having platform data.
> >> >>>>>>No, I think we're going for overkill here by not doing
> >> >>>>>>serial_pl01x.c as
> >> >>>>>>platform data.  ns16550 does platform data for this already.  This
> >> >>>>>>sounds like the lowest overhead way to get the clock
> >> >>>>>>populated and not
> >> >>>>>>have some DT data that's not going to be accepted upstream.
> >> >>>>>>
> >> >>>>>ummmm I am a bit lost at this point, could we recap please?
> >> >>>>Lets update the recap:
> >> >>>>- Please re-submit the dts file, now with whatever form is
> >> >>>>in v4.12-rc1,
> >> >>>>   saying as such in the commit (if it's just the commit message that
> >> >>>>   changes, that's fine and great).
> >> >>>The DTS file in v4.12-rc2 still does NOT contain the usb node.
> >> >>>
> >> >>>==> Should I then not use the DM on USB so I can avoid DTS changes?
> >> >>Well, you can either put it in the -u-boot.dtsi file for the board, and
> >> >>remove that later once it's upstream.
> >>
> >>
> >> yes I'll do that. thanks.
> >>
> >> >
> >> >>
> >> >>>>- Please update serial_pl01x.c to be able to get the clock
> >> >>>>via platform
> >> >>>>   data, update and test your board to confirm it works.
> >> >>>um, It gets tricky;
> >> >>>I can not use platform_data since I can not use SERIAL_DM because
> >> >>>the device tree does have a UART node which gets picked up.
> >> >>How about disabling the node in -u-boot.dtsi for the board and then you
> >> >>can use platform data,
> >> >
> >>
> >> I dont think that would because CONFIG_OF is enabled for USB; so the
> >> kernel dtsi that contains the uart node (without the clock!) will be
> >> picked by u-boot and the uart will not be initialized properly.
> >> I still think that the simplest solution is to let me merge with the
> >> kernel's device tree plus this u-boot.dtsi [1];
> >> then just get rid of the file when possible (and NEIHER the source
> >> code NOR the configs) would need to change
> >>
> >> [1] https://github.com/ldts/poplar-u-boot/blob/upstream/arch/arm/dts/hi3798cv200-u-boot.dtsi
> >
> > Yes, sorry.  [1] needs to be updated to disable uart0 so that you can
> > use platform data, at least for now.  I do want to talk more with Rob
> > about the general problem this exposes.
> 
> Using platform data because we cannot put a clock frequency in the DT
> node seems unfortunate to me. With a little flexibility, DT can be
> made to work. But IMO the sort of pedantry makes great the enemy of
> good.

This, and the alias issue we talked about the other day (wrt mmc) do
highlight what I see as problems of the dts being too kernel-centric.
But I also really want to wait for Rob to chime in before we get too far
here.

-- 
Tom
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 819 bytes
Desc: Digital signature
URL: <http://lists.denx.de/pipermail/u-boot/attachments/20170525/46d18186/attachment.sig>