[U-Boot] [PATCH 1/2] OMAP3 SPL: Rework memory initalization and devkit 8000 support

[Igor Grinberg]

On 10/27/2011 11:00 PM, Tom Rini wrote:
> On 10/27/2011 01:46 PM, Igor Grinberg wrote:
>> Hi Tom,
>>
>> That seems like a good change, though I would recommend to split it into
>> smaller
>> patches, please, see below:
>>
>> On 10/26/2011 11:13 PM, Tom Rini wrote:
>>> This changes to making the board be responsible for providing the
>>> memory initialization timings in SPL
>> That probably would be one patch.
>>
>>>  and converts the devkit 8000
>>> to this framework.
>> That, would be another one.
> But breaking it up like this breaks bisect'ability as devkit8000 builds
> today (well, outside of the scope of this series).

Well the rule for keeping bisectability was always:
1) add the new functionality
2) switch to it
3) remove the unneeded (old) one

And yes, you don't just move code to split the patch if you want
the bisectability kept, it could be more then that.

>>>   As part of this suffix the Micron DDR settings
>>> with their speed
>> Next patch.
> OK.
>
>>>  and add a few more timing values that will be needed.
>> This can go along with the patch that uses those settings.
> I'll double check where they're really neded.
>
>>> We also make sure that in mem_ok() we clear the values off as we may be
>>> testing the same banks multiple times.
>> That's should be another patch (although it has only one line).
> Sure.  Again, the bigger picture right now I'm hoping for is:
> (a) does it work on all of the other beagleboard revs?  The logic is a
> little twisty but it's what x-loader does today and (b) does the overall
> approach look sane to everyone?

a) Testing is always welcome.
    I don't have those boards, but there are many people who do.
b) Yes, I think it is pretty sane.

Regards,
Igor