[U-Boot] [PATCH v2 01/44] Correct defconfigs using savedefconfig

[Simon Glass]

Hi Tom,

On 6 September 2016 at 12:01, Tom Rini <trini at konsulko.com> wrote:
> On Mon, Sep 05, 2016 at 11:47:00AM +0900, Masahiro Yamada wrote:
>> 2016-08-30 9:21 GMT+09:00 Simon Glass <sjg at chromium.org>:
>> > Update the defconfig files to match their canonical form, as produced by
>> > 'make safedefconfig'.
>> >
>> > This is the result of running 'tools/moveconfig.py -s' on the tree.
>> >
>> > Signed-off-by: Simon Glass <sjg at chromium.org>
>>
>> >
>> > diff --git a/configs/10m50_defconfig b/configs/10m50_defconfig
>> > index 15952af..0e3ad96 100644
>> > --- a/configs/10m50_defconfig
>> > +++ b/configs/10m50_defconfig
>> > @@ -1,7 +1,7 @@
>> >  CONFIG_NIOS2=y
>> >  CONFIG_SYS_CONFIG_NAME="10m50_devboard"
>> > -CONFIG_DM_SERIAL=y
>> >  CONFIG_DM_GPIO=y
>> > +CONFIG_DM_SERIAL=y
>> >  CONFIG_DEFAULT_DEVICE_TREE="10m50_devboard"
>> >  CONFIG_FIT=y
>> >  CONFIG_OF_BOARD_SETUP=y
>> > diff --git a/configs/3c120_defconfig b/configs/3c120_defconfig
>> > index b19c956..0c3fbde 100644
>> > --- a/configs/3c120_defconfig
>> > +++ b/configs/3c120_defconfig
>> > @@ -1,7 +1,7 @@
>> >  CONFIG_NIOS2=y
>> >  CONFIG_SYS_CONFIG_NAME="3c120_devboard"
>> > -CONFIG_DM_SERIAL=y
>> >  CONFIG_DM_GPIO=y
>> > +CONFIG_DM_SERIAL=y
>> >  CONFIG_DEFAULT_DEVICE_TREE="3c120_devboard"
>> >  CONFIG_FIT=y
>> >  CONFIG_OF_BOARD_SETUP=y
>>
>>
>>
>> If the following patch is applied
>> http://patchwork.ozlabs.org/patch/664076/
>> the savedefconfig sync will produce completely different
>> results for these defconfigs.
>>
>>
>> The "make savedefconfig" sorts CONFIGs
>> in the order as Kconfig entries are parsed.
>>
>> In other words, the canonical order changes
>> every time bare default entries are added/removed
>> in board/*/Kconfig.
>>
>>
>> We are just repeating mad churn...
>
> Yes, but it's churn we have to live with until everything is migrated,
> no?

Yes, I think so. I originally put the TPL option down a bit in the
file and then moved it back when I saw the patch size.

We cannot expect the conversion to Kconfig to result in small patches
when a large number of boards are affected. I think we just need to
grin and bear it :-)

Regards,
Simon