[Question] Is little endian supported on all the platforms?

[Boqun Feng]

On Mon, Aug 31, 2015 at 09:19:26PM +1000, Michael Ellerman wrote:
> On Mon, 2015-08-31 at 15:53 +0800, Boqun Feng wrote:
> > On Mon, Aug 31, 2015 at 04:52:38PM +1000, Benjamin Herrenschmidt wrote:
> > > On Mon, 2015-08-31 at 14:44 +0800, Boqun Feng wrote:
> > > > Hi all,
> > > > 
> > > > I hit a strange build error on v4.2, when I try to build a LE kernel
> > > > with a slightly modification of the ppc64_defconfig. What I did is just
> > > > make ppc64_defconfig and make menuconfig to set CPU_LITTLE_ENDIAN=y, and
> > > > then build the kernel.
> > > > 
> > > > I did a little research myself, and found out the error is because of
> > > > trying to build a LE kernel with CONFIG_PPC_PS3=y. So a simple way to
> > > > reproduce is:
> > > 
> > > First, LE isn't supported on PS/3 :-) It's only supported on POWER8
> > > machines (and *some* P7 machines and only when running as KVM guest).
> > 
> > Got it, thank you ;-)
> > 
> > So should we modify Kconfigs of all the platform to reflect this?
> 
> Yes. Maybe something like this?
> 
> diff --git a/arch/powerpc/platforms/Kconfig.cputype b/arch/powerpc/platforms/Kconfig.cputype
> index c140e94..c359f72 100644
> --- a/arch/powerpc/platforms/Kconfig.cputype
> +++ b/arch/powerpc/platforms/Kconfig.cputype
> @@ -415,9 +415,13 @@ config VDSO32
>           big endian. That is because the only little endian configuration we
>           support is ppc64le which is 64-bit only.
>  
> +config CHOOSE_ENDIAN
> +       bool
> +
>  choice
>         prompt "Endianness selection"
>         default CPU_BIG_ENDIAN
> +       depends on CHOOSE_ENDIAN
>         help
>           This option selects whether a big endian or little endian kernel will
>           be built.
> diff --git a/arch/powerpc/platforms/powernv/Kconfig b/arch/powerpc/platforms/powernv/Kconfig
> index 604190c..32d5661 100644
> --- a/arch/powerpc/platforms/powernv/Kconfig
> +++ b/arch/powerpc/platforms/powernv/Kconfig
> @@ -18,6 +18,7 @@ config PPC_POWERNV
>         select CPU_FREQ_GOV_ONDEMAND
>         select CPU_FREQ_GOV_CONSERVATIVE
>         select PPC_DOORBELL
> +       select CHOOSE_ENDIAN

IIUC, we can actually set PPC_POWERNV=y and PPC_PS3=y at the same time,
right?

So this can't prevent PPC_PS3 && CPU_LITTLE_ENDIAN from being true.

>         default y
>  
>  config OPAL_PRD
> diff --git a/arch/powerpc/platforms/pseries/Kconfig b/arch/powerpc/platforms/pseries/Kconfig
> index 54c87d5..182f485 100644
> --- a/arch/powerpc/platforms/pseries/Kconfig
> +++ b/arch/powerpc/platforms/pseries/Kconfig
> @@ -21,6 +21,7 @@ config PPC_PSERIES
>         select HOTPLUG_CPU if SMP
>         select ARCH_RANDOM
>         select PPC_DOORBELL
> +       select CHOOSE_ENDIAN
>         default y
>  
>  config PPC_SPLPAR
> 
> 
> 
> > > But the specific problem you are having looks like a differnet issue
> > > with the PS3 boot wrapper.
> > > 
> > 
> > But could the reason of this be that the building process of the PS3
> > boot wrapper assumes the kernel is BE?
> 
> It looks more like we're getting confused between 32-bit and 64-bit, from your
> log:
> 
>     ld: powerpc:common64 architecture of input file `arch/powerpc/boot/ps3-head.o' is incompatible with powerpc:common output
> 
> Which says it has a .o which is 64-bit but it's trying to produce 32-bit output.
> 
> That's probably related to the PPC64_BOOT_WRAPPER stuff.

Probably, and just FYI, I build a 64bit BE kernel, the dtbImage.ps3 is
ELF 32-bit MSB executable.

Regards,
Boqun

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 473 bytes
Desc: not available
URL: <http://lists.ozlabs.org/pipermail/linuxppc-dev/attachments/20150831/73adf37a/attachment.sig>