[OE-core] [PATCH] u-boot: update to 2015.01

[Otavio Salvador]

On Thu, Jan 22, 2015 at 2:00 PM, Denys Dmytriyenko <denis at denix.org> wrote:
> On Thu, Jan 22, 2015 at 01:40:43PM -0200, Otavio Salvador wrote:
>> On Thu, Jan 22, 2015 at 1:17 PM, Denys Dmytriyenko <denis at denix.org> wrote:
>> > On Thu, Jan 22, 2015 at 03:02:06PM +0000, Burton, Ross wrote:
>> >> On 22 January 2015 at 14:57, Denys Dmytriyenko <denis at denix.org> wrote:
>> >>
>> >> > Please note that he's talking about current 2014.07 version, not the
>> >> > update!
>> >> > Can you confirm that it works or fails on the autobuilder?
>> >> >
>> >>
>> >> Sorry, got lost in the thread.  The old version works fine, the upgrade
>> >> breaks.
>> >
>> > That's not the case for the rest of us, unfortunately...
>> >
>> > I'm suspecting sstate got something to do with the fact that it still works
>> > for the old version on autobuilder.
>> >
>> >
>> > Let me emphasise it - this specific u-boot-fw-utils-cross recipe used to
>> > build fine, but now fails in the same exact way regardless of the version,
>> > either current 2014.07 or the updated 2015.01.
>> >
>> > So, it seems to me that fixing it is rather out of scope of this version
>> > update patch... :)
>> >
>> > I'm thinking of 2 short-term workarounds - don't update this particular recipe
>> > to the new version; or remove it altogether. I'm sure Otavio wouldn't be happy
>> > either way, but fixing it should probably be a separate effort...
>>
>> We should have denied the update on previous versions. It has been
>> broken by your update to 2014.07:
>>
>> Author: Denys Dmytriyenko <denys at ti.com>
>> Date:   Tue Sep 16 19:10:01 2014 -0400
>>
>>     u-boot: update to version 2014.07
>>
>>     u-boot doesn't really support building its tools for the target, as they are
>>     built with HOSTCC compiler, which is also used to compile fixdep utility
>>     that gets executed during the build. Since it might be beneficial to have a
>>     target version of mkimage, let's hack it to build fixdep in a separate step.
>>
>>     Signed-off-by: Denys Dmytriyenko <denys at ti.com>
>>     Signed-off-by: Ross Burton <ross.burton at intel.com>
>>     Signed-off-by: Richard Purdie <richard.purdie at linuxfoundation.org>
>>
>> and this has not been checked if it kept building at that time.
>
> Yes, it was building that time! And Ross has proof of that with autobuilder
> logs. He wouldn't have accepted it if autobuilder was breaking. It was
> breaking on another recipe and I fixed that. But fw-utils-cross was fine.
>
>
>> So I'm ok with this update to go and you work in fix it in a follow up
>> patch, but drop it is not acceptable. It has a clear use case.
>
> If it has such a clear use case, how do you use it when it's broken?

Well a non commonly used one? I didn't expect often use would be a
trade metric here but ...

-- 
Otavio Salvador                             O.S. Systems
http://www.ossystems.com.br        http://code.ossystems.com.br
Mobile: +55 (53) 9981-7854            Mobile: +1 (347) 903-9750