[U-Boot] [PATCH 8/8] Migrate generic bootcount to Kconfig

[Alex Kiernan]

On Mon, Feb 12, 2018 at 8:48 AM, Lukasz Majewski <lukma at denx.de> wrote:
> Hi Alex,
>
>> On Sun, Feb 11, 2018 at 9:44 PM, Lukasz Majewski <lukma at denx.de>
>> wrote:
>> > On Sun, 11 Feb 2018 21:04:46 +0000
>> > Alex Kiernan <alex.kiernan at gmail.com> wrote:
>> >
>> >>
>> >> That said, squashing in that change doesn't obviously break
>> >> anything for me, and is probably a step in the right direction.
>> >>
>> >> I'll see what Travis thinks.
>> >>
>> >
>> > We will probably receive build breaks...
>>
>> Yup... https://travis-ci.org/akiernan/u-boot/jobs/340344489
>>
>> Just tried again on one of those failures (x600) with the the default
>> removed and just set the on board that uses CONFIG_EXT, but that then
>> fails at config time.
>
> Ok. I see
>
>>
>> TBH I'd actually like to take it out of Kconfig (which I realise is
>> the wrong direction) as it just feels fundamentally wrong the way it
>> is.
>
> To finish moving SYS_BOOTCOUNT_ADDR to Kconfig we would need to add its
> definition to each eligible ./configs/<board>_defconfig file.
>
> Then we would have the previous behaviour preserved.
>
>> I don't know what the U-Boot approach configuration like this
>> is... struggling to find prior art.
>
> Let's ask Tom for his opinion as he did much such work before.
>

Thoughts:

- Do nothing, leave CONFIG_SYS_BOOTCOUNT_ADDR as a purely CONFIG_EXT
piece of Kconfig
- Rename CONFIG_SYS_BOOTCOUNT_ADDR to something like
CONFIG_SYS_BOOTCOUNT_ADDR_EXT for just CONFIG_EXT
- Remove CONFIG_SYS_BOOTCOUNT_ADDR from Kconfig altogether (and rename
to say SYS_BOOTCOUNT_ADDR)
- Pick through every config building defaults - okay for some boards,
but lots have it defined based on other memory offsets

I think the only real options are the last two.

Whatever we do, I think CONFIG_SYS_BOOTCOUNT_ADDR wants splitting into
at least two:

- I2C - an offset from an I2C base for the bootcounter
- Others - an actual address used for storing the bootcounter

I'm struggling to see why EXT is the way it is - AFAICS the location
it uses to access/return the bootcounter is basically local to the two
functions in bootcount_ext and could just be a local variable.

-- 
Alex Kiernan