On 15.11.18 20:33, Andy Shevchenko wrote:
> On Thu, Nov 15, 2018 at 8:28 PM Alexander Graf <ag...@suse.de> wrote:
>> On 15.11.18 18:58, Andy Shevchenko wrote:
>>> Cache the value of the reg-io-width property for the future use.
>
>>> struct ns16550_platdata {
>>> unsigned long base;
>>> + int reg_width;
>>> int reg_shift;
>>> - int clock;
>>> int reg_offset;
>>> + int clock;
>>
>> Why move clock?
>
> To group reg_* parameters together as it might make sense in the
> future to have some other struct which would be used inside
> serial_device_info and here.
> If it's a big deal, don't pay attention to it, I would not touch it by
> request.
Well, there are a few users of "patch" semantics:
* mainlining process
* patch review
* backports (to stable / distro versions)
* reverts if something broke
While I agree that you could ignore it for the normal mainlining
process, all other users will get hurt by having multiple "things" done
in a single patch. Review gets harder, backports get more confusing and
a revert ends up changing more code, which in turn makes it harder.
So please keep changes isolated to the thing they do :).
> P.S. Why to overquoute?
What do you mean? Why I don't remove useless parts of the quote?
Lazyness I suppose.
Alex
_______________________________________________
U-Boot mailing list
U-Boot@lists.denx.de
https://lists.denx.de/listinfo/u-boot
