On Fri, May 2, 2014 at 10:38 AM, Richard Purdie
<[email protected]> wrote:
> On Fri, 2014-05-02 at 10:34 +0200, Andrea Adami wrote:
>> I probably misunderstand that part.
>>
>> I could not see any difference in the produced klcc.cross scripts with
>> or without the mangling.
>> Tested runtime on 3 machines.
>> What I did not yet try is a rebuild from sstate but I expect it to be 
>> rebuilt.
>
> The easy way to reproduce a problem is to build for MACHINE=A and then
> build for MACHINE=B where the two machines have the same architecture.
> You will find that whilst it will switch the commented line at the top
> to refer to the new machine, the encoded lines do not get changed.
>

I did exactly build for c7x0 and poodle, two armv5te, and armv4 collie.
The 2nd line *and* the furter references to prefix et.al. were
apparently correct.
I'll retest.

> You would also find that if you run a build in dirA, then a build in
> dirB using the sstate cache from dirA, the paths would reference dirA,
> not dirB. After this change they will reference dirB.
>

Ah, I was fully unaware of that ! I'm not reusing the sstate cache.

> Cheers,
>
> Richard
>
Thanks

Andrea
-- 
_______________________________________________
Openembedded-devel mailing list
[email protected]
http://lists.openembedded.org/mailman/listinfo/openembedded-devel

Reply via email to