RFR(M): 8016586: PPC64 (part 3): basic changes for PPC64

[Vladimir Kozlov]

I will push shortly Part 3 changes into staging repo with corresponding 
renaming in closed sources. And I added next to Goetz's original patch:

make/linux/platform_ppc

@@ -2,11 +2,11 @@

  arch = ppc

-arch_model = ppc
+arch_model = ppc_32

  os_arch = linux_ppc

-os_arch_model = linux_ppc
+os_arch_model = linux_ppc_32

  lib_arch = ppc

Thanks,
Vladimir

On 6/15/13 3:00 AM, David Holmes wrote:
> On 14/06/2013 8:38 AM, Lindenmaier, Goetz wrote:
>> Hi,
>>
>> http://cr.openjdk.java.net/~goetz/webrevs/8016586-inlcudes/
>>
>> This change contains the #includes we need for the linuxppc64 port,
>> and some other minor changes.  It decides about the
>> file names in the cpu/ppc directory.
>>
>> We changed some of your #includes to use _32 extension, basically
>> where they are guarded by TARGET_ARCH_MODEL... .  I think it
>> would make sense to set TARGET_ARCH_MODEL_ppc_32 instead
>> of TARGET_ARCH_MODEL_ppc in your port, and name all files used
>> within that define with extension _32. In the webrev this is changed
>> consistently (not so in the patch).
>
> So this ppc32 is indeed spreading everywhere :(
>
> That is potentially going to require a whole bunch of changes in our
> makefiles, source file renaming and possibly even build file changes.
> This will take time to investigate.
>
> It would have been simpler to simply let ppc imply 32-bit and use ppc64
> as needed to distinguish. Our LIBARCH for this is likely to remain ppc
> not change to ppc32
>
> David
> -----
>
>> What are your politics about ordering the includes here? Most
>> other includes are alpha sorted, but the platforms here are not
>> at all. If you wish to, I'll establish a certain order.
>>
>> The changes in compileBroker.cpp and globals.hpp/EnableInvokeDynamic
>> are preliminary. They'll be reverted in later changes.
>>
>> I would be glad if you could review this change,
>>
>> Best regards & thanks,
>>    Goetz.
>>