[U-Boot] [U-Boot, 36/36] rockchip: add common board file for rockchip platform

[Dr. Philipp Tomsich]

> On 10 Apr 2018, at 14:32, Tom Rini <trini at konsulko.com> wrote:
> 
> On Tue, Apr 10, 2018 at 02:54:09PM +0800, Kever Yang wrote:
>> Hi Tom,
>> 
>> 
>> On 04/09/2018 06:35 AM, Tom Rini wrote:
>>>> I have do a lot of test and re-work in my local branch and at last make
>>>> it landed in
>>>> rockchip vendor U-Boot, with testing in most of SoCs(not including
>>>> rk3066/rk3188).
>>>> Well, I do try to split it into pieces, but I found that actually not
>>>> help very much
>>>> except waste much more time:
>>>> - The target is(very clear) to make rockchip soc board file in a good
>>>> shape with common files,
>>>>     instead of copy-paste for each soc(more than 10 of them now)
>>>> - then we need to identify what's common and what should go to soc and
>>>> board;
>>>> - remove using common rockchip timer and use arm generic timer instead
>>>> for armv7
>>>>     SoCs(rk3066 and rk3188 need still using rockchip timer)
>>>> - most soc need to do uart init, boot order select, and some
>>>> arch_cpu_init().
>>>> - don't break the boards already working, so I still leave some code
>>>> which not so common
>>>>     in board file, but I would like to remove or move them into right
>>>> place if I got a board
>>>>     to verify;
>>>> 
>>>> @Simon, @Tom,
>>>> This patch set is to remove some common files and add some other common
>>>> files for
>>>> all Rockchip SoCs, I have to make sure the whole patch set can running
>>>> good for all SoCs,
>>>> but it's really hard to make every patch to build and work perfect for
>>>> all SoCs, is there
>>>> any mandatory rules for this?
>>> So you mean possibly breaking some existing platforms?  I don't like the
>>> idea of doing that...
>> 
>> No, I'm not intent to to breaking some existing platforms,
>> this patch set including 36 patches, all the platform should work fine
>> after apply all these patches, but if only some of them applied,
>> there is compile error or running error because of feature missing.
> 
> OK.  Similar to the Linux kernel, it's not a good thing to break
> buildability of things during a patch series.


Independent of this: this is not a single series, but multiple series rolled
into one.  Once the commit messages are reworked to convey what’s
changed in a more meaningful way, this will be even more apparent
than it already is today.

Thanks,
Philipp.