[yocto-ab] Future of powerpc support in Yocto Project?

[Khem Raj]

Thanks Richard for putting this in front of AB

On 5/10/18 9:57 AM, Richard Purdie wrote:
> Right now we support four architectures equally, one of them being
> powerpc.
> 
> Unfortunately its status as a primary architecture is coming under
> threat as the support in gcc and glibc is at risk of being
> deprecated/dropped as its not being maintained in the same way as the
> other architectures are. Khem can likely provide more information as he
> has connections into those communities.

gcc 8.x has marked SPE port of ppc obsolete and can be enabled with 
--enable-obsolete, however it does mean that it will disappear in gcc 9+

https://github.com/gcc-mirror/gcc/blob/master/gcc/config.gcc#L236-L250

ppc 32-bit has been deprecated in many distros e.g. debian already for 
couple of years now from being a release architecture

the reference implementation for qemuppc uses a non-spe target so we are 
alive on that but we have done several fixes especially in toolchain 
area, extending upto core-metadata and packages to accommodate SPE

At this point we spend quite a bit of effort in keeping things rolling 
when it comes to ppc 32bit.

> 
> Right now we don't have any firm plans on what to do. What I do want to
> ask is whether any of the members have interest in powerpc, or whether
> we should consider dropping that support to focus on the others?
> 

Probably we can mark it secondary architecture which does not gate 
release and development, we can still keep accepting patches and fixes 
from community.

> Certainly the real hardware powerpc we have is elderly and not easily
> obtained any more. The qemu testing is less onerous but we have to as
> the question of whether we're focussing resources where we should be.
> 
> Any feedback on how people feel about powerpc support is welcome.
> 
> Cheers,
> 
> Richard
> 
> 
> 
>