[CRIU] [PATCH] build: simplify ppc64 builds

Laurent Dufour ldufour at linux.vnet.ibm.com
Tue Feb 23 08:30:11 PST 2016


On 23/02/2016 17:09, Cyrill Gorcunov wrote:
> On Mon, Feb 22, 2016 at 02:48:27PM +0100, Laurent Dufour wrote:
>> Since the move to nmk, zdtm failed to build on ppc64 because ARCH is
>> computed in the nmk and set to powerpc.
>> In fact it seems more efficient to set it to ppc64 when this is ppc64
>> or ppc64le since these 2 architecture should be handled the same way.
>>
>> Doing this fixes the zdtm build issue and simplify a bit the others
>> makefiles.
>>
>> Cc: Cyrill Gorcunov <gorcunov at gmail.com>
>> Signed-off-by: Laurent Dufour <ldufour at linux.vnet.ibm.com>
>> ---
>>  Makefile                       |  2 +-
>>  criu/Makefile                  |  3 +--
>>  scripts/nmk/scripts/include.mk |  2 +-
>>  test/zdtm/Makefile.inc         | 11 -----------
>>  4 files changed, 3 insertions(+), 15 deletions(-)
> 
> Laurent, look I would really love to stick with
> powerpc instead, that's how it goes in linux kernel
> which I prefer to follow on. That said I also think
> we might move arch/ppc64 to arch/powerpc. Would
> it be hard to go this way? Or there some strong
> arguments against? What do you think? Pavel?

I don't like "powerpc" which usually stands for Power PC 32 bits.

I think the reason the kernel is using "powerpc" is historical and there
are so many architecture managed beyond (ppc32, ppc64, embedded,
system...), that this may be confusing.

AFAIK, when dealing with user space packages, it is a common situation
to have 'ppc64' (and sometimes ppc64le) for PowerPC 64 bits and
'powerpc' for PowerPC 32 bits.

Furthermore this is following the uname -m returned value.




More information about the CRIU mailing list